Causal Analysis of Fish Kills in the Shenandoah and Potomac Rivers: 44 a Report from a Workshop Held January 16-18, 2007

1 NCEA-C-1971 2 April 2007 3

6 Causal Analysis of Fish Kills in

7 the Shenandoah and Potomac

8 Rivers: A Report from a

9 Workshop held

10 January 16-18, 2007

19 National Center for Environmental Assessment 20 Office of Research and Development 21 U.S. Environmental Protection Agency 22 Cincinnati, OH 45268 23 1 NOTICE 2 3 4 This report has been internally reviewed and is being distributed for informational 5 purposes only. It has not been formally released by the National Center for 6 Environmental Assessment and should not be construed to represent Agency policy. 7 Mention of trade names or commercial products does not constitute endorsement or 8 recommendation for use. This document is intended for internal Agency use only. 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 Preferred citation: 43 U.S. EPA. 2007. Causal Analysis of Fish Kills in the Shenandoah and Potomac Rivers: 44 A Report from a Workshop held January 16-18, 2007. U.S. Environmental Protection 45 Agency, National Center for Environmental Assessment, Cincinnati, OH. 46 NCEA-C-1971.

5/2/07 ii DRAFT: DO NOT CITE OR QUOTE 1 TABLE OF CONTENTS 2 3 4 Page 5 6 WORKSHOP PARTICIPANTS...... iv 7 8 1. INTRODUCTION ...... 1 9 10 2. THE CASES ...... 2 11 12 2.1. CASE 1—2006 FISH MORTALITIES IN LOWER NORTH FORK 13 SHENANDOAH RIVER, VA...... 2 14 2.2. CASE 2—2006 FISH MORTALITIES IN THE SOUTH BRANCH 15 POTOMAC RIVER, WV...... 4 16 17 3. CANDIDATE CAUSE DEFINITIONS...... 7 18 19 3.1. ANOXIA DUE TO LOW DISSOLVED OXYGEN...... 7 20 3.2. ANOXIA DUE TO GILL INJURY ...... 7 21 3.3. ANOXIA DUE TO LOW BLOOD OXYGEN AFFINITY...... 12 22 3.4. MORTALITY DUE TO OTHER PATHOGENIC MODES OF ACTION ..... 12 23 3.5. MORTALITY DUE TO HIGH Ph ...... 12 24 3.6. MORTALITY DUE TO pH FLUCTUATIONS...... 13 25 3.7. MORTALITY DUE TO HIGH AMMONIA CONCENTRATIONS ...... 13 26 3.8. MORTALITY DUE TO UNSPECIFIED TOXIC CHEMICALS ...... 13 27 3.9. MORTALITY DUE TO STARVATION...... 14 28 29 4. IDENTIFICATION OF PROBABLE CAUSES ...... 15 30 5. DATA NEEDS...... 23 31 6. REFERENCES...... 24 32 33 APPENDIX A: EVALUATION OF DATA...... 25 34 APPENDIX B: DATA NEEDS...... 89

5/2/07 iii DRAFT: DO NOT CITE OR QUOTE 1 WORKSHOP PARTICIPANTS 2

3 Name Organization E-mail 4 1. John Wirts WVDEP [email protected] 5 2. Andrew Johnson WVDEP [email protected] 6 3. Chris Barry WVDEP [email protected] 7 4. Patrick Campbell WVDEP [email protected] 8 5. Jim Hedrick WVDNR [email protected] 9 6. Laurie Olah WVDA [email protected] 10 7. Matt Monroe WVDA [email protected] 11 8. Doug Chambers USGS-WVWSC [email protected] 12 9. Stephen Reeser VDGIF [email protected] 13 10. Don Kain VADEQ [email protected] 14 11. Larry M. Simmons VADEQ [email protected] 15 12. Ted Turner VADEQ [email protected] 16 13. Stephen McIninch VCU [email protected] 17 14. Larry Merrill EPA Region III [email protected] 18 15. Amy Bergdale EPA Region III [email protected] 19 16. Frank Borsuk EPA Region III [email protected] 20 17. Lou Reynolds EPA Region III [email protected] 21 18. Kate Schofield USEPA [email protected] 22 19. Pat Shaw-Allen USEPA [email protected] 23 20. Sharon K. Taylor USEPA [email protected] 24 21. Susan Norton USEPA [email protected] 25 22. Glenn Suter USEPA [email protected] 26 23. Joel Allen USEPA [email protected] 27 24. Erin Quinlan USEPA [email protected] 28 25. Jim Lazorchak USEPA [email protected] 29 26. Molly Smith Friends of Shenandoah R. [email protected] 30 27. Karen Andersen Friends of Shenandoah R. [email protected] 31 28. Jeff Kelble Shenandoah River Keeper [email protected] 32 29. Jeroen Gerritsen Tetra Tech Inc. [email protected] 33 30. Don Catanzaro TN&A [email protected] 34 31. Lynnette Sholar TN&A [email protected] 35

5/2/07 iv DRAFT: DO NOT CITE OR QUOTE 1 1. INTRODUCTION 2 3 Mass mortalities of fish (kills) have occurred each year in the Shenandoah River 4 since 2002 and, to a lesser extent, in the South Branch of the Potomac River. A 5 workshop was convened by Region III of the U.S. Environmental Protection Agency 6 (U.S. EPA), the Virginia Department of Environmental Quality and the West Virginia 7 Department of Environmental Protection at Cacapon State Park, WV, January 16-18, 8 2007. The workshop addressed two kills that occurred in the spring of 2006. The goals 9 of the workshop were to apply the stressor identification process to those kills using 10 available data, to make a preliminary determination of the causes of the kills and to 11 develop a prioritized list of data needs for the 2007 season. This scope and set of goals 12 were determined during a preliminary workshop in September 2006. Participants in this 13 workshop included representatives of the States of Virginia and West Virginia, the U.S. 14 EPA, U.S. Geological Survey, and stakeholders and technical experts. 15 The method of analysis was the U.S. EPA’s stressor identification process, as 16 implemented in the CADDIS system (http://www.epa.gov/caddis). Stressor identification 17 involves five steps: define the case, list candidate causes, evaluate data from the case, 18 evaluate data from elsewhere and identify probable causes. It is a weight-of-evidence 19 approach in which all relevant and available data are collected, data are analyzed to 20 produce evidence of potentially causal relationships, the evidence is classified into 21 types, each type of evidence is scored, the results are summarized for each candidate 22 cause and, if possible, the probable cause is identified. The CADDIS process serves to 23 organize the available information for the candidate causes and present the evidence 24 for and against each one. 25 This report presents the results of the workshop, which narrowed the problem by 26 eliminating some candidate causes and identified data needs for future kills in the 27 Shenandoah or upper Potomac. Because essential data from field data sheets and 28 pathology reports were unavailable, the report does not demonstrate ability of CADDIS 29 to determine the causes of fish kills. However, it shows that even a preliminary 30 application of the CADDIS method can be useful.

5/2/07 1 DRAFT: DO NOT CITE OR QUOTE 1 2. THE CASES 2 3 2.1. CASE 1—2006 FISH MORTALITIES IN LOWER NORTH FORK 4 SHENANDOAH RIVER, VA 5 Case 1 is the fish mortality event that occurred between Woodstock and Passage 6 Creek on the North Fork Shenandoah River in Virginia during the spring of 2006 (Figure 7 1). Dead and dying fish were found along an approximately 30-mile stretch of the river 8 beginning in early March and continuing through late May of that year. A fish survey 9 performed 2 weeks prior to the initiation of the kill found that the fish were apparently 10 healthy. The species reported to be affected consisted primarily of smallmouth bass 11 (Micropterus dolomieu) and redbreast sunfish (Lepomis auritus); some dying 12 hogsuckers (Hypentelium roanokense) were also found. Adult fish appeared to be the 13 dominant age class affected, though some sub-adult smallmouth bass also died. 14 Biologists observed fin rot and other signs of stress in the dead fish and also in a small 15 number of rock bass (Ambloplites rupestris), green sunfish (Lepomis cyanellus), bluegill 16 (Lepomis macrochirus) and largemouth bass (Micropterus salmoides). Overall, 17 reproduction of smallmouth bass and recruitment of young stock appeared to be good 18 for the season. 19 Two distinct clinical phases appeared to occur during this mortality event on the 20 Lower North Fork. First, an acute phase in which large numbers of dead fish appeared 21 began shortly after the first dead fish were reported and lasted for several days around 22 March 20. The dying fish did not show any external lesions. Some fish were noted to 23 be lethargic and swimming at the surface of the water just prior to dying. In a second 24 phase, affected fish had skin lesions, eroded fins and gill hyperplasia (thickening of the 25 gill filaments). Several smallmouth bass that were sampled later that year exhibited 26 circular areas of regenerating scales that may indicate that some fish recovered from 27 the skin lesions. In addition, some smallmouth bass from these die-offs were found on 28 post-mortem examination to be “intersex” (males with eggs in their gonads). It is 29 unknown yet whether this condition is related to the fish deaths. Despite this intersex 30 condition, reproduction of the smallmouth bass in these waters was noted to be good.

5/2/07 2 DRAFT: DO NOT CITE OR QUOTE 1 2 FIGURE 1 3 Locations of the Reported 2006 Fish Kills in the South Branch of the Potomac River and 4 the North Fork of the Shenandoah River

5/2/07 3 DRAFT: DO NOT CITE OR QUOTE 1 The reference streams selected for the Virginia Shenandoah fish kill study are 2 Cedar Creek and the Cowpasture and Maury Rivers (Figure 2). Cedar Creek is a large 3 tributary of the North Fork Shenandoah River, and enters the North Fork just northwest 4 of the town of Strasburg. The Cowpasture River serves as an out-of-basin reference. It 5 and the Jackson River meet near Clifton Forge, VA to form the James River. The 6 Maury River is also an out of basin reference stream, and is a tributary of the James 7 River. The collection site on the Cowpasture River is located in the Walton Tract of the 8 George Washington National Forest, at the end of County Route 632 in Bath County. 9 The collection site selected for the Maury River is just downstream from a sewage 10 treatment plant (as are several sites on the North Fork Shenandoah River, and the 11 reference site on Cedar Creek) for the city of Lexington. Fish kills have not been 12 reported for any of these streams since the first kills were observed in the North Fork 13 Shenandoah River in 2004. All of these streams were selected as references because 14 they are similar to the Shenandoah North and South Fork Rivers in that they are 15 limestone buffered and agriculturally influenced. However, these influences are less 16 pronounced than those observed in the North and South Forks of the Shenandoah 17 River. 18 2.2. CASE 2—2006 FISH MORTALITIES IN THE SOUTH BRANCH POTOMAC 19 RIVER, WV 20 Case 2 includes fish mortalities that occurred in the South Branch of the Potomac 21 River Watershed in West Virginia during the late spring of 2006. The mortalities 22 extended from Moorefield and continued down stream for approximately 55 miles 23 (Figure 1). Multiple locations within this area were surveyed and found to have a few 24 dead fish at each location. Dead and dying fish were found beginning around May 25th 25 and continued through Memorial Day weekend. No additional dying or dead fish were 26 observed after May 31st of that year. The majority of dead fish were redhorse suckers 27 (Moxostoma spp.) along with some northern hogsuckers (Hypentelium nigricans) and a 28 few smallmouth bass (Micropterus dolomieu). Adult fish appeared to be the dominant 29 age class reported to be affected. Reproduction and recruitment was found to be good 30 for the season despite the kill.

5/2/07 4 DRAFT: DO NOT CITE OR QUOTE 1

2 FIGURE 2 3 Locations of Reference Sites for Comparison with the River Reaches in which the 2006 4 Fish Kills Occurred. Red areas are the watersheds of the South Branch of the Potomac 5 River and the North Fork of the Shenandoah River. For the South Branch, only the 6 primary reference site on the Greenbrier River is shown.

5/2/07 5 DRAFT: DO NOT CITE OR QUOTE 1 In this case, it appeared that the affected fish underwent only an acute phase in 2 which mortality was sudden. Fish were reported to be at the water’s surface, gasping 3 and behaving abnormally prior to dying. Upon examination, affected fish had gill 4 hyperplasia (thickening of the gill filaments) that diminished the surface area for oxygen 5 exchange. 6 Reference sites (no reported fish kills in spring 2006), both within basin and out 7 of basin, were set up for comparison of fish health and water chemistry data (Figure 2). 8 Continuous water quality monitors measure pH, dissolved oxygen (DO), conductivity, 9 and temperature at six reference sites; Patterson Creek, North Fork of the South 10 Branch, South Fork of the South Branch, North Branch of the Potomac, Opequon Creek 11 and Cacapon River. With the exception of the North Branch and Opequon Creek, these 12 sites are similarly dominated by agricultural impacts. The North Branch is a mining 13 impacted watershed and Opequon Creek is influenced heavily by municipal wastes. 14 The North Fork and South Fork monitoring sites are both upstream of the 2006 fish kills. 15 Smallmouth bass were collected for intersex and relative health determinations on the 16 Gauley River (at Camp Ceasar in Nicholas County), the Back Fork of the Elk River 17 (above Webster Springs) and the West Fork of the Greenbrier River (above Durbin) in 18 October of 2005. These three sites are considered as clean references due to their 19 wilderness nature.

5/2/07 6 DRAFT: DO NOT CITE OR QUOTE 1 3. CANDIDATE CAUSE DEFINITIONS 2 3 Nine candidate causes have been proposed for the North Fork Shenandoah and 4 South Branch Potomac fish kills. These candidate causes were chosen by participants 5 in a workshop held at Cacapon, WV, in September 2006. At that workshop, which 6 included stakeholders as well and academic, state and federal scientists, the available 7 evidence was presented, causal hypotheses were discussed and conceptual models of 8 the hypotheses were generated. All causal hypotheses that were still advocated by any 9 participant at the end of that workshop were included as candidate causes in the 10 January 2007 workshop. 11 The nine candidate causes include three different mechanisms for inducing 12 anoxia. A fourth is mortality due to pathogenic or parasitic infections acting through 13 mechanisms other than gill damage. These four candidate causes and the sources and 14 pathways that result in exposure of fish to those candidate causes are illustrated in a 15 single conceptual model (Figure 3). High pH levels, large pH fluctuations and high

16 ammonia (NH3) concentrations are other possible causes. Their sources and pathways 17 are illustrated in Figure 4. Another candidate cause is unspecified toxic chemicals 18 (Figure 5). The last candidate cause is starvation (Figure 6). 19 3.1. ANOXIA DUE TO LOW DISSOLVED OXYGEN 20 This candidate cause includes mortality from anoxia (extremely low blood oxygen 21 concentrations) due to low aqueous DO concentrations. Mortality in fish may occur 22 when DO levels fall below 4.0 mg/L. Low DO may result when nutrient enriched 23 streams produce large amounts of aquatic plant biomass. When the biomass 24 decomposes, bacteria consume available oxygen below levels required by fish for 25 survival. Low DO may also occur when point or non-point source discharges to streams 26 have a high biological or chemical oxygen demand. In addition, at night when 27 photosynthesis stops but respiration continues, algae contribute to oxygen depletion. 28 3.2. ANOXIA DUE TO GILL INJURY 29 Injuries to gills may result in insufficient gas exchange and death due to anoxia 30 even if aqueous DO concentrations are high. Injuries may result from mechanical 31 abrasion, parasitic or microbial infection, hyperplasia (thickening of the gill epithelium)

5/2/07 7 DRAFT: DO NOT CITE OR QUOTE Gro up 1 CM – O2 uptake & pathogens – 2/5/07

non -point sources point so urces

unidentified non- poultry farming golf rights- watershed lan d riparian la nd point sour ces wa stes cours es of-way industrial cover a lteration cover alteration stormwater effluents septic indust rial poultry pr ocess ing crop s overflows systems emissions wa stes WW TP detention effluents mi ning basins wa stes unidentified point so urces ↑ imper vious deicers surfaces fertilizers pesticides

↑ surf ace r unoff fish ↑ delivery of organic matter, nut rients, hatcheries & other substances t o stream chan nel ↑ eros ion stocked released alteration fish bait fish

precipitation

↑ fish pop ulation system ↑ phosphorus ↑ nitrogen ↑ end ocrine- ↑ me rcury ↑ other ↑ organic dens ity productivity disrupting c hemicals con taminants ma tter

↑ phot osynthesis temperature age of fish ↑ pathogens or parasites KEY source ↑ phys iological spawnin g ↓ immunocompetence stress modifying factor ↑ pH ↑ NH 3

additional step in causal pa thway ↑ suspe nde d solids - - ↑ NO 3 + NO 2 ↑ pathogenic or p arasitic infection proximate stressor

mode of action

biotic respons e ↑ gill injury (hyp erplasia or erosion) ↓ dissolved oxyge n temper ature

↑ methemogl obinem ia

↓ oxygen uptake

↑ fish mortality (smallmouth ba ss & suck ers)

1 2 FIGURE 3 3 Conceptual Model of Candidate Causes 1-4, all of which Result in Fish Mortality due to Low Blood Oxygen Levels

5/2/07 8 DRAFT: DO NOT CITE OR QUOTE Group 2 CM – ammon ia & pH – 2/6/07

non-point sources point sources

watershed la nd riparian land unidentified non- industrial poultry farming golf cove r alteration cove r alteration point sources emissions wastes cou rses industrial stormwater effluents septic crops overflows systems WW TP mi ning detention effluents basins poultry processing wastes unidentified ↑ impervious wa stes surfaces point sources fertilizers

↑ surface runoff ↑ delivery of organic matter, nutrients, & ot her substances t o stream chan nel ↑ erosion alteration

precipitation

↑ phosphorus ↑ nitrogen

flow light ↑ organic KEY ma tter source

modifying factor ↑ respiration ↑ photosynthesis

addit ional step in causal pathway

↓ pH ↑ pH ↑ NH temperature proximate stressor 3

mode of action

biotic response ↑ pH f luctuation ↓ NH 3 excretion

↑ internal NH phys iological 3 stress

↓ protein metabolism

↓ ion regulation ↑ gill hyp erplasia

↑ lesions ↑ gill or fin erosion ↑ lethargy

↑ fish mortality (smallmouth ba ss & suck ers) 1 2 FIGURE 4

3 Conceptual Model of Fish Mortality due to Candidate Causes 5-7, High pH Fluctuations, High pH Levels and High NH3 4 Levels

5/2/07 9 DRAFT: DO NOT CITE OR QUOTE Group 3 CM – unspecified toxics – 2/6/07

non-point sources point sources

unidentified non- exposed land poultry farming golf rights- point sources surfaces wastes courses of-way industrial stormwater effluents overflows septic poultry processing crops systems industrial wastes WWTP emissions detention effluents basins mining unidentified wastes point sources deicers

fertilizers pesticides

precipitation

↑ toxics in wet & dry deposition

↑ toxics adsorbed ↑ toxics in watershed land riparian land to particles cover alteration cover alteration discharged waters

↑ impervious ↑ toxics in ↑ toxics in surfaces subsurface waters surface runoff

↑ surface runoff ↑ unspecified toxics in stream ↑ delivery of toxics to stream (water or sediment) channel ↑ erosion alteration KEY source ↑ episodic ↑ sustained exposure to toxic exposure to toxic modifying factor

additional step in causal pathway

↑ bioaccumulation proximate stressor

mode of action

toxicity at sensitive times toxicity to sensitive toxicity to all biotic response (e.g., during fat metabolism) life stages life stages

↑ gill hyperplasia ↑ susceptibility to ↑ lethargy other stressors ↑ lesions ↑ gill or fin erosion

↑ fish mortality (smallmouth bass & suckers) 1 2 FIGURE 5 3 Conceptual Model of Candidate Cause 8, Unspecified Toxic Chemicals

5/2/07 10 DRAFT: DO NOT CITE OR QUOTE Group 4 CM – food resources – 2/6/07

non-point sources point sources

unidentified non- poultry farming golf rights- wastes watershed land riparian land point sources courses of-way industrial cover alteration cover alteration stormwater effluents septic industrial poultry processing crops overflows systems emissions wastes WWTP detention effluents mining basins wastes unidentified point sources ↑ impervious deicers surfaces fertilizers pesticides

↑ surface runoff fish ↑ delivery of organic matter, nutrients, hatcheries & other s ubstances to stream channel ↑ erosion stocked released alteration fish bait fish

precipitation

↑ phosphorus ↑ nitrogen ↑ endocrine- ↑ mercury ↑ other disrupting chemicals contaminants ↑ organic matter ↑ photosynthesis temperature system ↑ fish population productivity density

↑ invertebrate ↑ pH ↑ NH3 KEY pathogens & parasites source - - ↑ NO3 + NO2 modifying factor

additional step in light flow causal pathway

physical proximate stressor ∆ plant assemblages ↓ invertebrate prey habitat

mode of action

biotic response ↓ prey availability ↓ prey quantity ↓ prey quality

↑ physiological stress

↑ susceptibility to ↓ condition other stressors

↑ fish mortality (smallmouth bass & suckers)

1 2 FIGURE 6 3 A Conceptual Model of Candidate Cause 9, Starvation

5/2/07 11 DRAFT: DO NOT CITE OR QUOTE 1 and effects of toxicity. Any injuries that substantially reduce the ability of the gill to 2 extract oxygen from the water, through reductions in gill surface area or tissue 3 permeability to oxygen, may result in mortality. 4 3.3. ANOXIA DUE TO LOW BLOOD OXYGEN AFFINITY 5 Oxygen is exchanged from the water through the gill and is transported to fish 6 tissues via hemoglobin in red blood cells. Even if aqueous DO levels are high and gas 7 exchange by the gills is unimpaired, death may result from reduced ability of the blood 8 to carry oxygen. The most important cause is methemoglobinemia, a condition that 9 occurs in the presence of high concentrations of nitrites, in which the oxygen carrying 10 ferrous ion (Fe2+) of the heme group of the hemoglobin molecule is oxidized to the ferric 11 state (Fe3+), converting hemoglobin to methemoglobin, a non-oxygen binding form of 12 hemoglobin. High plasma pH may also result in reduced oxygen delivery to the tissues. 13 Severe cases of either condition may result in mortality. 14 3.4. MORTALITY DUE TO OTHER PATHOGENIC MODES OF ACTION 15 Mortalities may result from bacterial, viral, parasitic or fungal infections whenever 16 homeostatic functions in an organism are substantially compromised. In addition to 17 impairing gill function (candidate cause 2), pathogens can cause death by tissue 18 damage or wasting, septicemia, increased susceptibility to opportunistic secondary 19 infections, or other mechanisms. Heavy body burdens of parasites increase stress on 20 fish and render them more susceptible to infection and environmental perturbations, and 21 may result in mortalities during energy-intensive activities such as spawning and nest 22 building/ defense. Finally, some pathogens act by releasing toxins to the water. 23 3.5. MORTALITY DUE TO HIGH pH 24 High pH can directly kill fish by disrupting blood chemistry. Gill damage may 25 reduce delivery of oxygen to tissues and can reduce the effectiveness of gill ion 26 exchange and excretion. Compromised gill ion exchange can also potentiate the

27 toxicity of NH3 by reducing the ability of the fish to maintain sufficiently low plasma NH3 28 concentrations. Elevated ambient pH may also cause increased plasma pH, denaturing 29 proteins and contributing to tissue oxygen deficiency (Candidate Cause 3).

5/2/07 12 DRAFT: DO NOT CITE OR QUOTE 1 3.6. MORTALITY DUE TO pH FLUCTUATIONS 2 Sufficiently high pH can result in mortality, even if the duration at those levels is 3 brief. Rivers frequently exhibit diel fluctuations resulting from photosynthesis. As plants 4 photosynthesize during the day, they sequester dissolved carbon dioxide, resulting in 5 elevated pH. However, at night photosynthesis ceases but plants continue to consume 6 DO and release carbon dioxide, which lowers pH. Typical fluctuations during sunny 7 days under low-flow conditions may cause pH to range from 8.0 to 9.3 or greater; in 8 some instances hydronium ion concentration may vary by as much as a factor of 100. 9 Extremely high concentrations may cause similar damages as observed for sustained 10 high pH, and substantial fluctuation can challenge the fish’s ability to maintain 11 homeostasis. 12 3.7. MORTALITY DUE TO HIGH AMMONIA CONCENTRATIONS

13 High concentrations of aqueous NH3 may kill fish, independently of their potential

14 contributions to gill injury and anoxia (Candidate Cause 2). The toxicity of NH3 is a

15 function of the un-ionized concentration. The fraction of un-ionized NH3 is determined 16 primarily by the pH and temperature of the water. In general, the higher the pH, the

17 greater the total fraction of un-ionized NH3. Toxic concentrations of un-ionized NH3 are 18 temperature- and species-specific, and susceptibility of fish varies with age. Stress 19 during the reproductive season or when coming out of winter or high blood ammonia

20 due to protein metabolism may also lower fish tolerance to NH3. 21 3.8. MORTALITY DUE TO UNSPECIFIED TOXIC CHEMICALS 22 The chemical composition of the rivers, particularly during the periods of the fish 23 kills, is not well known. It is plausible that some episodic increase in exposure to an 24 unknown chemical or combination of chemicals could have caused the kills. Both rivers 25 receive input from point sources including WWTPs, industrial discharges and non-point 26 sources including urban runoff. These sources release agricultural and residential 27 nutrients, pesticides, herbicides and other chemicals. Atmospheric deposition may play 28 a substantial role in toxicant inputs, such as in the case of mercury and acid rain. In 29 sufficient concentrations, toxicants from any of these diverse sources may result in fish 30 mortalities. Observed patterns of mortality will vary depending on the source of the 31 input.

5/2/07 13 DRAFT: DO NOT CITE OR QUOTE 1 3.9. MORTALITY DUE TO STARVATION 2 Starvation may occur due to lack of food, inability of the fish to capture or ingest 3 food or inability of the fish to assimilate nutrition from ingested materials. In general, 4 health indices such as length to weight ratios give an acceptable indication of whether 5 starvation is occurring.

5/2/07 14 DRAFT: DO NOT CITE OR QUOTE 1 4. IDENTIFICATION OF PROBABLE CAUSES 2 3 Because the field and laboratory data sheets for fish collected during both kills 4 were not available at the time of the workshop in January 2007, it was not possible to 5 identify the probable cause of either case. The consistency of the available evidence 6 and the information gaps can be identified by examining scoring Tables 1 and 2. 7 For each candidate cause, those of us attending the workshop analyzed the data 8 that were available data to produce evidence which was evaluated in terms a system of 9 scores applied to each type of evidence. Details of the analysis of evidence and scoring 10 can be found in Appendix A. 11 The following scores were applied to the evidence. 12 +++ Convincingly supports 13 --- Convincingly weakens 14 ++ Strongly supports 15 -- Strongly weakens 16 + Somewhat supports 17 - Somewhat weakens 18 0 Ambiguous 19 NE No Evidence 20 P Pending evidence 21 R Refutes.

5/2/07 15 DRAFT: DO NOT CITE OR QUOTE TABLE 1

Summary of Evidence Concerning the 2006 Fish Kill in the North Fork of the Shenandoah River

Anoxia due to Mortality due to Low Blood Toxicity from Unspecified Starvation from Anoxia due to Anoxia due to Gill Other Toxicity from pH Toxicity from South Branch Oxygen Affinity or NH (in the Toxic Low Food Low DO Injury or Hyperplasia Pathogenic Fluctuations High pH 3 Methemoglo- environment) Substances Resources Modes binemia Type of Notes Score Notes Score Notes Score Notes Score Notes Score Notes Score Notes Score Notes Score Notes Score Evidence Evidence that uses data from the case

Spatial/ DO R Pathology NE(P) Pathol- -/ Pathol- -/ Kill did - - - Kill did - - - NH3 not - - - Little NE Dead --- Temporal Co- was data are ogy data NE(P) ogy data NE(P) not co- not co- elevated analy- fish were occurrence not low missing. are nega-tive occur occur at time sis of not during missing. or with with of kill toxics notice- the kill missing. highest high pH ably fluctu- starved ations Evidence of Charac- 0 Charac- 0 Behavior 0 NE NE NE No NE No NE Exposure or teristic teristic undiag- useful stomach Biological behaviors behavior nostic; tissue content Mechanism but multiple but analyse data causes multiple s causes

Causal pH 0; NH3 0; + Sources 0 Some + Potential + Poten- + Poten- + Poten- + No food NE Pathway Temp +; possible, possible sources tial tial tial data Autointox nitrite sources are sources sources sources cycle +; unknown, of patho- present are are are nitrate -; other gens present present present nitrite -; TSS mecha- -; conduct -; nisms seasonal possible stress +; behavioral stress +; flow -; diatoms - Laboratory Tests of NE Tests of Site post-kill Media water Verified Predictions

5/2/07 16 DRAFT: DO NOT CITE OR QUOTE 1 TABLE 1 cont.

Anoxia due to Low Blood Oxygen Mortality due to Toxicity from NH Unspecified Starvation from Anoxia due to Anoxia due to Gill Toxicity from pH Toxicity from 3 South Branch Affinity or Other Pathogenic (in the Toxic Low Food Low DO Injury or Hyperplasia Fluctuations High pH Methemoglo- Modes environment) Substances Resources binemia Type of Notes Score Notes Score Notes Score Notes Score Notes Score Notes Score Notes Score Notes Score Notes Score Evidence Symptoms Possibly +(P) Pathol-ogy NE(P) Pathol- NE(P) Pathol- NE(P) Pathol- NE(P) Pathol- NE(P) Pathol- NE(P) No NE consistent, data are ogy data ogy data ogy data ogy data ogy data pathology data are missing. are are are are are data missing. missing. missing. missing. missing. missing. Evidence that uses data from elsewhere Stressor- Similar - Compari 0 Local ref. - Response fluctuatio son to sites are Relationships ns non- previous not from other lethal year elevated Field Studies

Stressor- Some +(P) Cen- --- pH not - NH3 -- Response evidence trarchids at lethal levels too Relationships suggestive are levels low from but more resistant during Laboratory analysis kills Studies needed Analogous 2002 kill may + 2005 kill 0 Cases be may be analogous analogo but data us anecdotal or unavailable Considerations Based on Multiple Lines of Evidence Consistency Co- NA Consistent + Inconsiste - Few, - Negative - Lack of - Evidence --- Evi- 0 Only - occur- but few and ncies weak and except co- consist- dence evidence rence weak lines of inconsist for occur- ently too is alone evidence ent data sources rence negative, meager negative refutes but some ambiguity

5/2/07 17 DRAFT: DO NOT CITE OR QUOTE TABLE 1 cont.

5/2/07 18 DRAFT: DO NOT CITE OR QUOTE 1 TABLE 2

Summary of Evidence Concerning the 2006 Fish Kill in the South Branch of the Potomac River

5/2/07 19 DRAFT: DO NOT CITE OR QUOTE 1 TABLE 2 cont.

Anoxia due to Mortality due to Low Blood Toxicity from NH Unspecified Starvation from Anoxia due to Anoxia due to Gill Other Toxicity from pH Toxicity from 3 North Fork Oxygen Affinity (in the Toxic Low Food Low DO Injury Pathogenic Fluctuations High pH or Methemoglo- environment) Substances Resources Modes binemia Type of Notes Score Notes Score Notes Score Notes Score Notes Score Notes Score Notes Score Notes Score Notes Score Evidence Causal H2O pH +; pH + Sources 0 Sources + Potential + Potential + Potential + Poten- + Food NE Pathway fluctuations +; of of path- sources sources sources tial avail- H2O NH3 -; precusor ogens were were were sources ability H2O temp +; s are were present present present were unknow autointox cycle known present present n +; nitrate -; but not nitrite +; TSS -; the conduct -; pathway spawning/ prespawning -; seasonal hab changes -; flow +; planktonic diatoms 0; periphytic diatoms -; wash of dead diatoms - Laboratory Tests of NE Tests of Site water Media after the kill Temporal Sequence Verified Predictions Symptoms Pathology data NE(P) Patholog NE(P) Pathol- NE(P) Patholog NE(P) Patholog NE(P) Pathology NE(P) Pathol- NE(P) Pathol- NE(P) are missing y data ogy data y data y data data are ogy ogy are are are are missing data data are missing missing missing missing are missing missing

5/2/07 20 DRAFT: DO NOT CITE OR QUOTE TABLE 2 cont.

Anoxia due to Mortality due to Low Blood Toxicity from NH Unspecified Starvation from Anoxia due to Anoxia due to Gill Other Toxicity from pH Toxicity from 3 North Fork Oxygen Affinity (in the Toxic Low Food Low DO Injury Pathogenic Fluctuations High pH or Methemoglo- environment) Substances Resources Modes binemia Type of Notes Score Notes Score Notes Score Notes Score Notes Score Notes Score Notes Score Notes Score Notes Score Evidence Evidence that uses data from elsewhere Stressor- Similar - Similar - No NE Response fluturation NH3 did relevant Relationships s non- not cause analy- from other lethal kills ses Field Studies Stressor- Some +(P) Ambient - - NE No NE Field pH - NH3 levels -- Response evidence concentr relevant non- too low Relationships suggestive but ations lab lethal in from more analysis are too studies lab but Laboratory needed low found other Studies species Analogous 2002 kill is NE No data NE Parasites + Data from NE Data NE Data from NE Data NE Data NE Cases analogous but from ob- prior kills from prior prior kills from from data were prior kills served in lacking kills lacking prior prior unavailable 2002 kill lacking kills kills lacking lacking Considerations Based on Multiple Lines of Evidence Consistency Only NA Ambiguous 0 Inconsis- -/0 Positive + Few but - Few but - Consis- -- Evi- NE All data -- co- tent and but weak consis- consis- tent and dence weaken, occur- weak tent tent negative too but rence meager meager is needed Reasonable Nitrite - Sources - Sources - Sources - Explanation of elevated are not are not are not the Evidence but not sufficient sufficient sufficient enough 1

5/2/07 21 DRAFT: DO NOT CITE OR QUOTE 1 The workshop participants reached the following conclusions. 2 3 • Anoxia as a result of low DO was not a possible cause in either case 4 (Sections A.1.1 and A.2.1). 5 • Starvation was highly unlikely in both cases because the fish did not 6 appear starved and the condition of surviving fish was similar to that of fish 7 at reference locations (Sections A.1.9 and A.2.9).

8 • Elevated NH3 and pH levels and pH fluctuations also appear not to be 9 causes of either kill. Levels do not appear to have been elevated at the 10 time of either kill relative to times when no kills occurred, and they do not 11 appear to have been sufficiently elevated to cause the kills. However, 12 they may have contributed to the susceptibility of fish to other agents. 13 • The evidence for anoxia due to low blood oxygen affinity or 14 methemoglobinemia was strongly negative for the Potomac because 15 measured nitrite concentrations were low. Evidence was strongly 16 negative for the Shenandoah because the species that were killed are 17 resistant to that candidate cause. 18 • Data were insufficient to evaluate unspecified pathogens or toxic 19 chemicals as causes. Either one is plausible but both are largely 20 unsupported. 21 • Anoxia due to gill injury is supported by anecdotal reports of apparently 22 injured gills and by the occurrence of agents that could cause gill injury 23 (although not necessarily at sufficient levels). Hence, the evidence is 24 weak and it is not clear which of the causal pathways would be 25 responsible. 26 27 In summary, of the 9 candidate causes, 6 are unsupported or unlikely given the 28 evidence. Anoxia due to gill injury, pathogens and toxins remain as candidate causes. 29 Some additional analysis of data from these two cases and of data from other laboratory 30 and field studies could improve the causal analysis. However, the greatest 31 improvement would come from high quality field and laboratory pathology data. Data 32 gaps identified during the workshop are summarized in the following section and listed 33 in Appendix B.

5/2/07 22 DRAFT: DO NOT CITE OR QUOTE 1 5. DATA NEEDS 2 3 Workshop participants identified areas where additional data would reduce 4 uncertainty and provide additional types of evidence. In general, taking measurements 5 and collecting samples at the same time and place of the observed effect and at least 6 one matched reference site was the most important aspect of sampling. Appendix B 7 gives the outcome of this brainstorming session, which outlines data needs in three 8 lists. The first list is of data needs by general categories of information. The second list 9 provides specific needs for two of the candidate causes. The third list gives the highest 10 priority needs as identified by individual workshop participants. 11 The data gap receiving the most attention at the workshop was a baseline of fish 12 health; that is, what does a healthy fish in these rivers look like? When we expand this 13 question by considering the number of species involved—smallmouth bass, redbreast 14 sunfish, rockbass, golden redhorse and northern hogsucker—combined with the large 15 numbers of parameters that could be measured, the question gets very complex. 16 However, Appendix B represents a good starting point for data needs that should be re- 17 visited and revised as needs arise and are filled.

5/2/07 23 DRAFT: DO NOT CITE OR QUOTE 1 6. REFERENCES 2 3 Animal Disease Diagnostic Laboratory. 1998. Nitrite Toxicosis in Freshwater Fish 4 "Brown Blood Disease." Purdue University.

5 Constable, M., M. Charlton, F. Jensen, K. McDonald, G. Craig and K.W. Taylor. 2003. 6 An ecological risk assessment of ammonia in the aquatic environment. Human Ecol. 7 Risk Assess. 9:527-548.

8 Gillevet, P.M., P. Bojo, M. Sikaroodi, G. Garman, and S. McIninch. Preliminary LH-PCR 9 Analysis of Mucosal Surfaces of Smallmouth Bass from the Shenandoah River, 10 Unpublished report, George Mason University.

11 Kahn, C.M., Ed. 2005. Merck Veterinary Manual., 9th ed. Merck & Co., Inc., 12 Whitehouse Station, NJ.

13 Kroupova, H., J. Machova and Z. Svobodova. 2005. Nitrite influence on fish: A review. 14 Vet. Med. 50:461-471.

15 Milne, I., J. Seager, M. Mallett and I. Sims. 2000. Effects of short-term pulsed 16 ammonia exposure on fish. Environ. Toxicol. Chem. 19:2929-2936.

17 Schäperclaus, W. 1991. Fish Diseases, Vol.2. Amerind Publishing Co. Pvt. Ltd., New 18 Delhi 19 20 Scott, D.M., M.C. Lucas and R.W. Wilson. 2005. The effect of high pH on ion balance, 21 nitrogen excretion and behaviour in freshwater fish from an eutrophic lake: A laboratory 22 and field study. Aqu. Toxicol. 73:31-43.

23 Serafy, J.E. and R.M. Harrell. 1993. Behavioral response of fishes to increasing pH 24 and dissolved oxygen - field and laboratory observations. Freshwat. Biol. 30:53-61.

25 Smutna, M., L. Vorlova and Z. Svobodova. 2002. Pathobiochemistry of ammonia in the 26 internal environment of fish (review). Acta. Vet. Brno. 17:169-171.

27 Stoskopf, M.K. 1993. Fish Medicine. W.B. Saunders Company, Montreal.

28 U.S. EPA. 1999. Update of Ambient Water Quality Criteria for Ammonia. U.S. 29 Environmental Protection Agency, Office of Research and Development, Office of 30 Water, Washington, DC. EPA-822-R-99-014.

31 Wood, C.M. 2001. Toxic responses of the gill. In: Target Organ Toxicity in Marine and 32 Freshwater Teleosts: Vol. 1 -- Organs. D. Schlenk and W.H. Benson, Ed. Taylor & 33 Francis, New York, NY. 89 pp.

5/2/07 24 DRAFT: DO NOT CITE OR QUOTE 1 APPENDIX A 2 EVALUATION OF DATA 3 4 This appendix presents the evidence that was available at the workshop for each 5 candidate cause with respect to the North Fork Shenandoah case and then the South 6 Branch Potomac case. For each candidate cause, the evidence is analyzed in terms of 7 the types of evidence for which data were available. Data that were not available but 8 may exist are noted. 9 A.1. CASE 1—2006 LOWER NORTH FORK SHENANDOAH RIVER, VA 10 A.1.1. Anoxia due to Low Dissolved Oxygen. 11 A.1.1.1. Spatial/Temporal Co-occurrence — 12 Data Used Data from USGS continuous YSI DO monitors are not available for the fish 13 mortality event due to instrument failure. 14 There are temporal data from a reference site on the South Fork Shenandoah River. 15 Ad hoc measurements were taken during the kill. 16 17 Analysis Co-occurrence inferred from observations of dying fish and concurrent ad hoc 18 DO measurements of 8 mg/L by Steve Reeser. 19 20 Discussion The co-occurrence of dying fish and normal DO levels refutes the possibility 21 that low DO caused anoxia. Mortality from low DO is rapid. Therefore, if DO 22 concentrations are high but fish are still dying, low DO cannot be the cause. 23 24 Score R, These data are sufficient to refute low dissolved oxygen as the cause of the 25 kill. 26 27 A.1.2. Anoxia due to Gill Injury. 28 A.1.2.1. Spatial/Temporal Co-occurrence — 29 Data Used Gross and histopathology samples and field data from the kill and from 30 reference sites on the South Fork Shenandoah and Cowpasture Rivers were taken. 31 However, those data were not provided for this analysis. 32

5/2/07 25 DRAFT: DO NOT CITE OR QUOTE 1 Analysis None 2 3 Discussion The USGS verbally reported gill hyperplasia but data are unavailable. 4 5 Score NE (P) No evidence is available from the pathology reports, which are pending. 6 7 A.1.2.2. Evidence of Exposure or Biological Mechanism — 8 Data Used Observations of affected fish reported as lethargy and decreased flight 9 behavior response, but documentation was unavailable. 10 11 Analysis None 12 13 Discussion Lethargy and decreased flight response are common to many causes of 14 death and, therefore, are diagnostically non-specific. 15 16 Score 0 Ambiguous because the evidence also applies to other causes. 17 18 A.1.2.3. Causal Pathway —

19 Data Used Water pH (including pH fluctuations), water temperature, water NH3 levels, 20 water nitrite/nitrate levels and total suspended solids data. 21 Pathogen and parasite data may become available from samples sent to the pathology 22 lab. 23

24 Analysis Analyses of pH and NH3 data are presented in Sections A.1.5 through A.1.7. 25 Analyses of pathogen occurrence await the data sheets. 26 27 Discussion Gill injury could have been caused by several causal pathways. Evidence 28 for occurrence of a causal pathway consists of the occurrence of an agent in a pathway 29 leading to the proximate cause. Some of the agents in the pathway to this candidate

30 cause are also candidate causes themselves (i.e., pH, NH3 and pathogens). However, 31 the evidential requirements here are different. Elevated concentrations of an agent that

5/2/07 26 DRAFT: DO NOT CITE OR QUOTE 1 is a proximate cause must occur at the time of a kill, but agents in a causal pathway 2 may have been elevated in the past. It is necessary only that their effects continue. 3 pH - Data showed wide pH fluctuations on a daily basis, pH levels rose to 9.5. Although 4 large pH fluctuations did not consistently co-occur with the kill (Section A.1.6), they 5 may have contributed to gill injury and stress to the fish through alkalosis or 6 ammonia autointoxication.

7 NH3 - It was not elevated at the time of the kill (Section A.1.7), but it could have 8 contributed to the susceptibility of fish to another proximate cause of gill injury and 9 anoxia. Toxic gill necrosis can result from autointoxication (metabolism of proteins 10 in excess of excretory capacity) which may be triggered by starvation, a sudden drop 11 in temperature, an increase in pH above 9, an increase in aqueous ammonia, or an 12 increase in protein in the diet (Smutna et al., 2002). Of these, only elevated pH was 13 observed (see above).

14 Nitrate and Nitrite - Data were unavailable, but may be elevated by nitrification of NH3. 15 Temperature - Water temperature rose at the time of the kill, which increases biological 16 activity, including oxygen requirements and toxic responses.

17 Auto-intoxication - The elevated NH3, pH and temperature suggest that conditions could

18 have existed for an NH3 auto intoxication cycle to be initiated internally in the fish. 19 Conductance - Conductance was not elevated (USGS provisional data, Strasburg gage, 20 2006). 21 Pathogens - Data were available from analyses of bacterial and protist communities in 22 mucus from fish with and without lesions collected during the mortality event 23 (Gillevet et al., 2006). Parasite levels were potentially increased, but pathology data 24 were unavailable. 25 General stress - The kill did not correspond to any particular phase in the spawning 26 cycle. Seasonal behavior and habitat changes were noted that year; the fish did not 27 make typical habitat selection and movements apparently because of unusually low 28 water flow. 29 Abrasion - Concentrations of total suspended solids were not high. Periphytic, but not 30 planktonic, diatoms were quite heavy resulting in observations of “white rocks,” but

5/2/07 27 DRAFT: DO NOT CITE OR QUOTE 1 there were no high flows at the time of the kill to wash off diatom frustules and cause 2 abrasion. 3 4 Score Water pH 0 5 Water NH3 0 6 Water Nitrate & Nitrite NE 7 Water Temperature + 8 Auto intoxication cycle + 9 Conductance - 10 Stress, Spawning Season - 11 Stress, Season behavior and habitat changes + 12 Abrasion, Total Suspended Solids - 13 Abrasion, Flow - 14 Abrasion, Planktonic diatoms - 15 Abrasion, Wash off of dead periphytic diatoms - 16 17 A.1.2.4. Stressor-Response from Laboratory Studies —

18 Data Used Toxicologic benchmarks from laboratory tests of fish for pH, NH3 and 19 temperature. 20 pH 21 pH 9 causes carp mortality (Schaperclause, 1952) 22 When water pH > blood pH, NH3 excretion is reduced; above pH 9.5 it is blocked 23 (Schaperclause, 1952; Wood, 2001). 24 Other effects of high pH at the gills include increased CO2 excretion leading to 25 alkalosis and reduced uptake of sodium (Na) and chloride (Cl) ions (Wood, 26 2001). 27 Reported gill injuries were limited to hyperplasia of the chloride cells, not gross 28 hyperplasia (Wood, 2001). 29 30 Temperature 31 Rapid changes in temperature can increase morbidity and mortality in fish 32 (Stoskopf, 1993). 33 34 NH3 35 0.2 mg/L caused gill injury in brown trout (U.S. EPA, 1999). 36

37 Gill lamellae obtained from parental fish exposed to un-ionized NH3

38 concentrations ranging from 0.02 mg NH3-N/L to 0.05 mg NH3-N/L for 4 months, and

39 0.05 mg NH3-N/L and 0.06 mg NH3-N/L for 7 and 11 months, showed mild to moderate

5/2/07 28 DRAFT: DO NOT CITE OR QUOTE 1 fusion, aneurysms and separation of the epithelia from the underlying basement 2 membrane (U.S. EPA, 1999). 3 “In contrast to acute exposures, a variety of morphologic changes in the gills 4 have been described during chronic sublethal ammonia exposure. Most prominent are 5 an overall swelling of the respiratory lamellae, proliferation of epithelial cells, increased 6 diffusion distance and an increased prevalence of bacterial gill disease… These 7 responses would be expected to decrease the respiratory gas exchange capacity of the 8 fish, and thus its swimming performance and tolerance to hypoxia” (Wood, 2001). 9 Smallmouth bass are fairly sensitive. The growth effect concentrations ranged

10 from 0.05 mg NH3-N/L at pH=6.6 to 0.71 mg NH3-N/L at pH=8.68 (U.S. EPA, 1999, pg 11 118). 12 Redhorse sensitivities are unknown

13 As fish emerge from torpor in the spring, toxic NH3 concentrations drop from 1.7 14 to 0.2 mg/L (Schaperclause, 1952).

15 0.5 mg/L total NH3 highest observed in spring. 16 17 Nitrite and Nitrate 18 Chronic exposures to nitrite can cause gill hyperplasia (Kroupova et al., 2005). 19 20 Analysis Data were not available at the workshop to evaluate the potential for observed 21 levels of stressors to injure gills. Benchmark values should be compared to 22 concentrations measured at the time of the kill and shortly before. Also differences in 23 sensitivity in the laboratory should be compared to apparent differences in response 24 among species during the kill. 25 26 Discussion NA. 27 28 Score +, (P) Some evidence supports the candidate cause but more is pending 29

5/2/07 29 DRAFT: DO NOT CITE OR QUOTE 1 A.1.2.5. Analogous Cases — 2 Data Used Reports of state biologists. 3 Water quality data are potentially available from prior kill events. 4 5 Analysis Comparison of potentially analogous kills 6 7 Discussion Two similar kills of smallmouth bass have occurred in the Shenandoah 8 River previously: 9 North Fork 2004 spring mortality event 10 South Fork 2005 spring mortality event 11 These kills differ from the 2006 kill in that they occurred directly after a major run-off 12 event while the 2006 kill did not. The 2004 and 2005 kills were more widespread. Also, 13 the water was extremely warmed in 2004 and 2005, but the warming was less in 2006.

14 In 2005, the smell of NH3 from a terrestrial source was quite strong. Also in 2005, white 15 suckers displayed gill hyperplasia, trematodes and parasites and skin lesions. Some 16 water chemistry data and water temperatures are potentially available for those kills. 17 Analysis of these data may provide information relevant to the 2006 kill. 18 19 Score +, Reports of gill injury in 2005 weakly support the candidate cause. 20 21 A.1.3. Anoxia due to Low Blood Oxygen Affinity—(Methemoglobinemia) 22 A.1.3.1. Spatial/Temporal Co-occurrence — 23 Data Used Plasma samples and data from the fish mortality event and reference sites 24 on the South Fork Shenandoah and Cowpasture Rivers are potentially available. 25 Observations of blood by State biologists recounted at the workshop. 26 27 Analysis Pending 28 29 Discussion Ideally, co-occurrence would be established by signs of 30 methemoglobinemia in fish from the kill but not elsewhere. State biologists observed

5/2/07 30 DRAFT: DO NOT CITE OR QUOTE 1 neither brown gills nor blood in these fish. High pH can also cause low blood affinity 2 without brown blood. Data from samples sent to the pathology lab were not available. 3 4 Score -/NE (P) Field observations were negative but more definitive data are needed 5 from the pathology reports. 6 7 A.1.3.2. Evidence of Exposure or Biological Mechanism — 8 Data Used See A.1.2.2. 9 10 Analysis See A.1.2.2. 11 12 Discussion See A.1.2.2, behavioral evidence is consistent with many causes. 13 14 Score 0 Ambiguous 15 16 A.1.3.3. Causal Pathway — 17 Data Used Knowledge of sources and processes in the watershed. 18 19 Analysis None. 20

21 Discussion Nitrite accumulates in water when NH3 concentrations are elevated and the

22 second stage of nitrification is inhibited. Sources of NH3 are present in the watershed, 23 but nitrification rates or processes controlling the rates are unknown. Nitrite 24 concentrations were not available at the workshop. 25 26 Score 0 Ambiguous because sources are known but the conversion processes are not. 27 28 A.1.3.4. Stressor-Response from Laboratory Studies — 29 Data Used Literature reviews. 30 Centrarchids (includes sunfish and black bass) are refractory to methemoglobinemia 31 (Kahn, 2005).

5/2/07 31 DRAFT: DO NOT CITE OR QUOTE 1 Methemoglobinemia symptoms occur at 0.10 to 0.50 mg/L in sensitive species (channel 2 catfish and trout) and LC50 values range from 0.60 to 200 mg/L (Animal Disease 3 Diagnostic Laboratory, 1998). 4 5 Analysis Logic and comparison to ambient concentrations. 6 7 Discussion If methemoglobinemia was the cause of the kill, Centrarchids would have 8 been among the last rather than the first to die. 9 10 Score - - This evidence greatly weakens methemoglobinemia as a candidate cause. 11 12 A.1.4. Mortality due to Other Pathogenic Modes of Action. 13 A.1.4.1. Spatial/Temporal Co-occurrence — 14 Data Used Gross and histopathology samples and field data from the mortality event 15 and from reference sites on the South Fork Shenandoah and Cowpasture River were 16 taken, but the data were not available for this analysis. 17 Data were available from analyses of bacterial and protest communities in livers from 18 fish with and without lesions collected during the mortality event (Gillevet et al., 2006). 19 20 Analysis Pending for pathology data. 21 Authors’ reported results for Gillevet et al. (2006). 22 23 Discussion The USGS verbally reported occurrences of parasites at the September 24 preliminary workshop, but data are unavailable. Gillevet et al. (2006) reported no 25 differences in bacteria and protists from livers. 26 27 Score -/NE(P) Results from Gillevet et al. were negative but not definitive. The 28 pathology reports are necessary to confirm or refute this candidate cause. 29

5/2/07 32 DRAFT: DO NOT CITE OR QUOTE 1 A.1.4.2. Evidence of Exposure or Biological Mechanism — 2 Data Used Observational data on affected fish demonstrating lethargy and decreased 3 flight response. 4 5 Analysis Minimal analysis due to non-specific observational nature of data. 6 7 Discussion Observational data in this case are diagnostically non-specific. 8 9 Score 0 Ambiguous, because the behavios have multiple causes. 10 11 A.1.4.3. Complete Exposure Pathway — 12 Data Used Knowledge of State biologists. 13 14 Analysis None. 15 16 Discussion Sources of pathogens may include stocked fish, released bait fish or 17 effluents from hatcheries. Trout but not smallmouth bass are stocked in the 18 Shenandoah. Bait minnows inevitably are released. Hatcheries occur at and below the 19 kill site. 20 21 Score + Somewhat supports because some steps are present. 22 23 A.1.4.4. Analogous Cases — 24 Data Used Reports of State biologists. 25 26 Analysis None. 27 28 Discussion In the 2005 Shenandoah River fish kill event (Section A.1.2.5), white 29 suckers displayed trematodes and other parasites and skin lesions. 30 31 Score + Reports of parasites in 2005 somewhat support the candidate cause.

5/2/07 33 DRAFT: DO NOT CITE OR QUOTE 1 A.1.5. Mortality due to High pH. 2 A.1.5.1. Spatial/Temporal Co-occurrence — 3 Data Used Continuously monitored pH data from 2005-2006 on the North Fork of the 4 Shenandoah at Woodstock and Strasburg sites collected by the USGS, Ken Hyer 5 ([email protected]). 6 7 Analysis Daily maxima were plotted with respect to time and the period of the kill. 8 Plot of pH over time with respect to the interval of the kill. 9 10.0 North Fork Shenandoah, Strasburg 10 [2006 USGS data (provisional)] m

11 u m

i 9.0 12 x ma

13 H

p 8.0 y

14 il a D 15 7.0 16 1/23 3/14 5/3 6/22 8/11 9/30 11/19 17 Date 18 19 FIGURE A-1 20 The daily maximum pH values in the North Fork Shenandoah River at Strasburg. The 21 grey band covers the period of the kill and the vertical line denotes the peak of the acute 22 portion of the kill. 23 24 25 Discussion pH values were high (above 9) beginning as early as February, however, 26 there was no kill at this time (Figure A-1). The kill did not begin until mid March, when 27 pH values were actually dropping. The highest pH of the recorded period came in 28 middle to late May, which was still during the timeframe of the kill, but by this point, fish 29 had been dying for 2 months, so this elevation alone wouldn’t have been the cause of 30 the kill. All this evidence demonstrates a lack of temporal alignment of peaks in pH and 31 the onset of the fish kill in the North Fork. 32

5/2/07 34 DRAFT: DO NOT CITE OR QUOTE 1 Score --- The effect both does not occur when the candidate cause occurs and does 2 occur when the candidate cause does not occur. 3 4 A.1.5.2. Complete Exposure Pathway — 5 Data Used Observations of State biologists. 6 7 Analysis None. 8 9 Discussion High pH in the Shenandoah results from the karst geology of the valley and, 10 during the day, from high plant production. 11 12 Score + Somewhat supports because sources are present. 13 14 A.1.5.3. Stressor-Response Relationship in the Lab — 15 Data Used Scott et al. (2005) addresses effects of pH as high as 9.5 on behavior of 16 rainbow trout and perch. 17 An article by Serafy and Harrell (1993) addressing sublethal effects of high pH on 18 bluegill, striped bass and killifish. 19 20 Analysis Maximum pH values during the kill were compared with those causing effects 21 in these studies. 22 23 Discussion The pH maximum during the study by Scott et al. (2005) was as high as 24 9.5, somewhat higher than observed on the North Fork in 2006 during the kill. Since 25 fish exposed to that pH were shown to have reduced ability to excrete ammonia (as 26 measured in the fish’s blood), but exhibited no lethality or change in swimming behavior, 27 the fish in this case should not have been dying due to similarly high pH levels alone. 28 Note that this experiment was carried out on different species. 29 30 The fish in the Serafy and Harrell (1993) experiment were subjected to pH increases of 31 about 1 unit over the course of less than 1 hour, with the highest replicate reaching a

5/2/07 35 DRAFT: DO NOT CITE OR QUOTE 1 final pH of about 9.5. Because the exposures are so short in duration, the comparability 2 with our case is severely reduced. Also, while these results may be useful for 3 assessing fish stress response, they are not very useful for assessing lethality since 4 none of the fish died. 5 6 Score - The fish exposed to pH levels similar to those found in our case showed 7 reduced ammonia excretion capabilities, but did not die. 8 9 A.1.5.3. Stressor-Response Relationship in the Field — 10 Data Used Continuous pH data from 2005-2006 on the North Fork of the Shenandoah 11 at Woodstock and Strasburg sites collected by the USGS, Ken Hyer 12 ([email protected]). 13 14 Analysis Maximum pH values from 2005 of 8.9 at Strasburg and 8.3 at Woodstock 15 were selected from this period and compared to maximum values during the case. 16 17 Discussion We compared the pH maximum during the analogous kill period from 2005 18 to the maximum during the kill of 2006. Because there were reports of lesions from the 19 North Fork in 2005 but very few kills, we might consider this a non-kill year. However, 20 the status of the 2005 scenario as a kill is confounded by the fact that there was a large 21 kill on the North Fork in 2004, and Virginia biologists hypothesize that the lack of a 22 large-scale kill in 2005 may simply have been a result of overall lower numbers of fish 23 and thus lower numbers available to die. 24 25 Score 0 The ambiguity of 2005 as a kill/non-kill year prevents us from drawing any 26 conclusions from the comparison of maximum pH levels. 27

5/2/07 36 DRAFT: DO NOT CITE OR QUOTE 1 A.1.6. Mortality due to pH Fluctuations. 2 A.1.6.1. Spatial/Temporal Co-occurrence — 3 Data Used Continuous pH data from 2005-2006 on the North Fork of the Shenandoah 4 at Woodstock and Strasburg sites collected by the USGS, Ken Hyer 5 ([email protected]). 6 7 Analysis Calculated pH shifts over 24 hour windows were plotted against the date and 8 with respect to the kill interval. 9 10 1.0 North Fork Shenandoah, Strasburg [2006 USGS data (provisional)] n

11 io t a u 12 t c u l 0.5 13 f H 14 p ily a

15 D 0.0 16 1/23 3/14 5/3 6/22 8/11 9/30 11/19 17 Date 18 19 FIGURE A-2 20 The Daily Difference Between Maximum and Minimum pH Values in the North Fork 21 Shenandoah River at Strasburg. The grey band covers the period of the kill and the 22 vertical line denotes the peak of the acute portion of the kill. 23 24 25 Discussion pH fluctuations were high (0.6-0.8 units) before, during and after the kill, 26 with the highest fluctuations coming at the end of the kill (Figure A-2). pH fluctuations 27 dropped to a low of only 0.1-0.2 units in April/May and the kill did not cease. Also, since 28 data begins only in February and the kill started in March, there are few preceding data 29 to establish a trend. Though there may be some interaction/combination of temperature 30 and pH effects on the fish, when considering pH fluctuations alone, there is not 31 significant evidence of temporal co-occurrence to support pH fluctuations as a 32 candidate cause.

5/2/07 37 DRAFT: DO NOT CITE OR QUOTE 1 Score --- The effect both does not occur when the candidate cause occurs and does 2 occur when the candidate cause does not occur. 3 4 A.1.6.2. Complete Exposure Pathway — 5 Data Used Knowledge of State biologists. 6 7 Analysis None. 8 9 Discussion The source of high pH fluctuations is the high primary production during the 10 day, which raises pH. High production is in turn due to high nutrient levels. 11 12 Score + Somewhat supports because some steps are present. 13 14 A.1.6.3. Stressor-Response Relationship in the Field — 15 Data Used Serafy and Harrell (1993) addressed sublethal effects of pH fluctuations on 16 natural fish communities including pumpkinseed, eel, and killifish. 17 18 Analysis We compared fish behavior in our case with that observed in the field 19 experiment under similar pH conditions. 20 21 Discussion The fish communities in this experiment were monitored during a natural 22 shift in pH through the course of a day while pH was monitored. The observers noted 23 shifts in community density, biomass and species richness. None of these changed 24 drastically through the course of the pH shift. pH was measured as varying from ~8.5 25 up to a maximum of ~9.8 and then down to a minimum of ~7.6. The pH regime in this 26 experiment is probably the most relevant external scenario to ours because of the 27 magnitude of pH shift and the high average pH around which the pH was centered. On 28 the other hand, the other environmental/ecological factors in this setup decreased 29 comparability between the experimental scenario and ours on the North Fork. The 30 study was conducted in a tidal freshwater environment of the Chesapeake Bay with 31 different fish species. We question whether, if this pH regime is natural and common for

5/2/07 38 DRAFT: DO NOT CITE OR QUOTE 1 that area, then are the native fish more accustomed and better adapted to these 2 conditions? Also, the observers noted whether or not fish fled the study site, which was 3 a hydrilla grass bed, and interpreted this as a potential reaction to undesirable pH 4 levels. But if pH remained the same at distances further from the grass bed, we 5 shouldn’t expect fleeing the grass bed to be of any benefit, so perhaps fleeing is an 6 inappropriate endpoint of assessment. Also, parameters other than pH that weren’t 7 measured may have influenced the fishes’ behavior. 8 9 Score The experiment somewhat weakens this candidate cause by showing that pHs 10 fluctuations of over 2 units centered around a pH of ~8.5, similar to the fluctuations 11 measured in our case, are not lethal to fish. However, the key differences of conditions 12 in each scenario reduce their comparability. 13 14 A.1.6.4. Stressor-Response Relationship in the Field — 15 Data Used Continuous pH data from January-October 2006 on the Rappahannock 16 River at gage #1668000 collected by the USGS, Ken Hyer ([email protected]). 17 18 Analysis Calculated daily pH shift plotted against time. 19 3.0 20 Rappahannock River [2006 USGS data (provisional)] n io

21 t a

u 2.0 t

22 c u l f

23 H

p 1.0

24 ily a 25 D 0.0 26 1/3 2/22 4/13 6/2 7/22 9/10 10/30 27 Date 28 29 FIGURE A-3 30 The Daily Difference Between Maximum and Minimum pH Values in the Rappahannock 31 River. The grey band covers the period of the kill in the Shenandoah River and the 32 vertical line denotes the peak of the acute portion of that kill.

5/2/07 39 DRAFT: DO NOT CITE OR QUOTE 1 Discussion The Rappahannock River is being used here as an out-of-basin reference 2 stream, though it should be noted that there is no solid evidence of similarities in 3 geology, ecology, and water quality between this reference stream and the North Fork. 4 That being said, these data show pH fluctuations of even greater magnitude (1.5-2 5 units, Figure A-3) than those seen on the North Fork (Figure A-2), and there were no 6 reported kills there. 7 8 Score - A difference in response relative to exposure to the candidate cause was 9 observed at non-spatially linked sites, but the difference is not in the expected direction. 10 The Rappahannock shows greater pH fluctuation than the South Branch but with no 11 increase in fish kill occurrence (in fact, no observed fish kills). This was not given a 12 stronger negative score because of the weak similarities in conditions between the kill 13 site and this reference site. 14 15 A.1.7. Mortality due to High Ammonia Concentrations. 16 A.1.7.1. Spatial/Temporal Co-occurrence — 17 Data Used Ammonia samples taken at an average of 2.5 days per week for sites at 18 north and south-run extremities and 5 days per week for sites in between north and 19 south sample runs at random days of the week from March-July 2006 on the North Fork 20 of the Shenandoah at Woodstock, Strasburg and Mount Jackson sites as well as South 21 Fork sites, Cedar Creek, Cowpasture River and Maury River at Bean's Bottom and 22 upstream of Mill Creek sites. This was a continuation of a similar weekly sampling 23 regime in 2005 except that the 2005 project included only North Fork Shenandoah sites 24 from Timberville downstream to Strasburg. Sampling has continued since July 2006 at 25 weekly intervals. pH readings were taken with these samples to allow for calculation of 26 un-ionized ammonia. Data was collected by the Valley Regional Office of the Virginia 27 DEQ and is in the possession of Robert (Ted) Turner, biologist for the DEQ. 28 29 Analysis We calculated un-ionized ammonia concentrations from the total ammonia 30 measurements using temperature and pH measurements taken at the time of sampling. 31 Then we plotted both un-ionized and total ammonia concentrations against time 32 (Figures A-4 and A-5).

5/2/07 40 DRAFT: DO NOT CITE OR QUOTE 1 0.16

) North Fork Shenandoah

2 /L [2006 VA DEQ data] Strasburg

3 mg 0.12 Woodstock N

( Mt. Jackson 3

4 H

d N 0.08

5 e v ol

6 s s 0.04 di 7 l a

8 Tot 0.00 9 2/17 4/8 5/28 7/17 Date 10 11 FIGURE A-4 12 Total Dissolved Ammonia Values in the North Fork Shenandoah River at Three 13 Locations. The Strasburg and Woodstock locations are within the kill zone, but the Mt. 14 Jackson location is an upstream reference site. The grey band covers the period of the 15 kill and the vertical line denotes the peak of the acute portion of the kill. 16 17 0.010 North Fork Shenandoah 18 [2006 VA DEQ data] ) Strasburg /L 19 Woodstock mg

20 N Mt. Jackson ( 3 21 0.005 NH d e

22 z i n o 23 i Un 24 0.000 25 2/17 4/8 5/28 7/17 Date 26 27 FIGURE A-5 28 Un-ionized Ammonia Values in the North Fork Shenandoah River at Three Locations. 29 The Strasburg and Woodstock locations are within the kill zone, but the Mt. Jackson 30 location is an upstream reference site. The grey band covers the period of the kill and 31 the vertical line denotes the peak of the acute portion of the kill.

5/2/07 41 DRAFT: DO NOT CITE OR QUOTE 1 Discussion Because un-ionized ammonia we calculated its concentrations. The 2 amount of un-ionized ammonia increases with pH. Also, since pH fluctuates through 3 the day, if ammonia is sampled in the morning and pH is recorded then, we expect pH 4 to rise by the afternoon, with a concurrent rise in un-ionized ammonia concentration. 5 Therefore, we may not be recording the day’s maximum un-ionized ammonia value if 6 we record in the morning. 7 8 Score --- The effect both does not occur when the candidate cause occurs, and does 9 occur when the candidate cause does not occur. 10 11 A.1.7.2. Complete Exposure Pathway — 12 Data Used Knowledge of State biologists. 13 14 Analysis None. 15 16 Discussion Sources of ammonia include sewage treatment plants, poultry wastes, and 17 nitrogenous fertilizers. 18 19 Score + Somewhat supports because some steps are present. 20 21 A.1.7.3. Stressor-Response Relationship in the Field — 22 Data Used Measurements of ammonia at similar sites both in the same basin and out- 23 of-basin in Virginia from March-October 2006. 24 Rivers/sites sampled include the North Fork of the Shenandoah at the Mt. Jackson 25 gauging station (upstream of the zone with reported kills), Cedar Creek, the Cowpasture 26 River, the Maury River at Bean’s Bottom and the Maury River upstream of Mill Creek. 27 28 Analysis Using total ammonia measurements, pH and temperature data, we calculated 29 un-ionized ammonia concentrations. We then plotted those values against time 30 (Figures A-4 and A-5). The Mt. Jackson data is plotted on the same graph presented 31 above alongside the data from sites within the kill zone. In addition, the other reference

5/2/07 42 DRAFT: DO NOT CITE OR QUOTE 1 streams are plotted in the same manner (Figures A-6 and A-7). The first pair of plots 2 shows total ammonia and the second pair, un-ionized ammonia. 3 4 0.16 North Fork Shenandoah - reference sites ) [2006 VA DEQ data] 5 /L

6 mg 0.12 N Cedar ( 3

7 H Cowpasture Maury @ Bean's Bottom d N 0.08

8 e

v Maury @ Mill Creek ol

9 s s 0.04 di 10 l a

11 Tot 0.00 12 2/17 4/8 5/28 7/17 Date 13 14 FIGURE A-6 15 Total Ammonia Values in Four Reference Streams for the North Fork Shenandoah 16 River. The grey band covers the period of the kill and the vertical line denotes the peak 17 of the acute portion of the kill. 18 19 0.010 North Fork Shenandoah - reference sites [2006 VA DEQ data] 20 ) 21 /L Cedar

mg Cowpasture N

22 (

3 Maury @ Bean's Bottom 23 H 0.005 Maury @ Mill Creek d N e 24 z oni i

25 n U

26 0.000 27 2/17 4/8 5/28 7/17 Date 28 29 FIGURE A-7 30 The Daily Un-ionized Ammonia Values in Four Reference Streams for the North Fork 31 Shenandoah River. The grey band covers the period of the kill and the vertical line 32 denotes the peak of the acute portion of the kill.

5/2/07 43 DRAFT: DO NOT CITE OR QUOTE 1 Discussion While the data recorded at the site of the kills does show a peak in the 2 concentration of ammonia in mid March, around the time of the first observed kills, we 3 also see a corresponding increase upstream of where kills were reported at the Mt. 4 Jackson gauging station. If ammonia alone were responsible for killing the fish, we 5 would expect to see kills everywhere we see these elevated ammonia levels, yet we do 6 not. Furthermore, ammonia levels drop again after the spike, yet fish continued to die. 7 If ammonia concentration falls to levels similar to those in the reference streams with no 8 observed kills, we would expect a cessation in the fish kill on the North Fork. As a final 9 caveat, we have not considered time of day of sampling here, and since pH fluctuates 10 throughout the day, it is possible that the variation we observe in these plots is partially 11 due to choice of sampling time. 12 13 Score - We observed a qualitative difference in response relative to exposure to the 14 candidate cause, at reference sites, but the difference was not in the expected direction. 15 The reference streams here are non-spatially linked sites and show an increase in 16 ammonia concentration similar to that in the North Fork but with no increase in fish kill 17 occurrence (in fact, no observed fish kills). 18 19 A.1.7.4. Stressor-Response Relationship in the Field 20 Data Used Monitored natural ammonia levels on the Rappahannock are frequently as 21 high as 0.6-0.9 mg/L. There have not been any reported kills there. This anecdotal 22 evidence comes from studies by Steve McIninch, Virginia Commonwealth University. 23 24 Analysis None, data are pending. 25 26 Discussion This evidence was available only as a verbal communication. 27 28 Score - (P) We saw a large difference in response relative to exposure to the candidate 29 cause at a reference site, but the difference is not in the expected direction. The 30 reference stream shows higher ammonia concentrations than the South Branch but no 31 occurrence of fish kills.

5/2/07 44 DRAFT: DO NOT CITE OR QUOTE 1 2 A.1.7.4. Stressor-Response Relationships in the Field — 3 Data Used Ammonia measurements covering the analogous kill period of 2005 4 recorded on the North Fork at Strasburg and Woodstock. 5 6 Analysis We calculated maximum un-ionized ammonia values of 7.3 µg/L at Strasburg 7 and 3.8 µg/L at Woodstock from this period. 8 9 Discussion The un-ionized maximum ammonia level during the analogous kill period 10 from 2005 is compared here to the max during the kill of 2006 under consideration. 11 Because there were reports of lesions from the North Fork in 2005 but few deaths, we 12 might consider this a non-kill year. However, the status of the 2005 scenario as a kill is 13 confounded by the fact that there was a large kill on the North Fork in 2004, and Virginia 14 biologists hypothesize that the lack of a large-scale kill in 2005 may simply have been a 15 result of overall lower numbers of fish and thus lower numbers available to die. So, 16 while ammonia levels were lower in 2005, we cannot draw any conclusions since we 17 don’t know whether to consider this a kill. 18 19 Score 0 The ambiguity of 2005 as a kill/non-kill year prevents us from drawing any 20 conclusions from the comparison of maximum un-ionized ammonia levels. 21 22 A.1.7.5. Stressor-Response Relationship from the Lab — 23 Data Used Articles including Milne et al. (2000) and Constable et al. (2003) and the 24 U.S. EPA (1999) criteria document. 25 Exposure data come from ammonia levels reported above from the Strasburg and 26 Woodstock sites on the North Fork. 27 28 Analysis We compared results reported in literature to ammonia levels reported above 29 from the Strasburg and Woodstock sites on the North Fork, above. 30

5/2/07 45 DRAFT: DO NOT CITE OR QUOTE 1 Discussion Reported ammonia thresholds for lethal or even sublethal toxic effects were 2 at least 10 times higher than what we’ve seen in the North Fork. For example, Milne et 3 al. (2000) reported needing ~400 µg/L un-ionized ammonia to show significant lethality 4 over a 24-hour exposure, ~750 µg/L over a 6-hour exposure, and ~850 µg/L for a 5 1-hour exposure in rainbow trout. In longer term pulsed exposure experiments of up to 6 6 weeks in duration, ~200 µg/L of un-ionized ammonia was needed to cause sub-lethal 7 effects assessed as “very severe gill damage.” Milne also noted the trend that 8 increased exposure frequency is more significant than increased total concentration of

9 ammonia in terms of effects on fish. Constable et al. (2003) reported an LC50 of ~1.3 10 mg/L of un-ionized ammonia for acute exposure in fathead minnows. TheU.S. EPA

11 (1999) criteria document reports values of EC20 of 4.79 mg/L of total ammonia at pH 8 12 for white sucker, 1.35 and 2.8 mg/L ammonia for bluegill and a geometric mean of 4.56

13 mg/L total ammonia at pH 8 for smallmouth bass. The document also reports an LC50 14 of 50 mg/L of total ammonia during an acute exposure test in fathead minnow. Highest 15 values in the North Fork were about 1-5 µg/L of un-ionized ammonia during the kill, 16 reaching a maximum for the sampling period of ~8 µg/L of un-ionized ammonia but only 17 at the end of the kill. It is definitely worth noting that none of these lab tests exposed 18 the fish to elevated or pulsed levels of ammonia for as long a period as our the fish in 19 this case may have been exposed. Indeed, it would be impractical to reproduce those 20 conditions. However, we’re also comparing the peak concentrations of ammonia 21 observed on the North Fork with reported lethal and sub-lethal levels from the literature. 22 So even if these observed levels approach those shown to cause effects in fish in the 23 literature, we know that ammonia levels fall far below this in between peaks and so the 24 average ammonia concentration the fish are exposed to as a sustained exposure is 25 considerably lower and definitely comparable with observed levels in reference streams 26 without kills. 27 28 Score -- If ammonia alone were responsible for the kills, we would expect the levels 29 recorded in the North Fork to be more similar to those shown to have killed fish in lab 30 tests, and that’s not the case. 31

5/2/07 46 DRAFT: DO NOT CITE OR QUOTE 1 2 A.1.8. Candidate Cause: Unspecified Toxics. 3 A.1.8.1. Evidence from the Case: Spatial-Temporal Co-occurrence — 4 Data Used Sediment, metals/organics and water chemistry data from two sites on the 5 North Fork of the Shenandoah. Virginia Probabilistic Monitoring Program (ProbMON)

6 VADEQ daily and continuous water column monitoring measures nutrients and NH4 but 7 no metals or organics. 8 Data from Friends of the Shenandoah. 9 10 Analysis NA. 11 12 Discussion Barium (Ba) was detected in the water column in 2006. No other data were 13 available for 2006 on toxic metals, organics, pesticides, pharmaceuticals. 14 Table A-1 presents the results of analyses of various media collected in the North Fork. 15 The utility of the data is limited, because they are not temporally and spatially relevant 16 to the fish kill events. The table should, however, be reviewed to determine similarities 17 and differences between the two watersheds to help design future studies by identifying 18 contaminants that should be monitored. 19 20 Score NE No Evidence. 21 22 A.1.8.2. Evidence from the Case: Evidence of Exposure or Biological 23 Mechanisms — 24 Data Used Arsenic (As) was observed in fish tissue data in 2005 (the year before); 25 2006 fish tissue data not yet available. 26 USGS Virginia laboratory (Paul McCormick) is investigating periphyton at 14 sites on 27 the NF Shenandoah. Data are pending. 28

5/2/07 47 DRAFT: DO NOT CITE OR QUOTE 1 Table A-1

Chemicals Detected in Various Media at Virginia Department of Environmental Quality (North Fork Shenandoah) and USGS (South Branch Potomac) Sample Stations

North Fork Shenandoah (river South Branch Potomac miles) Spring South Pilgrim's At Run Small Fork Mainstem Near Near Pride Peters- Moore- Peters- At Hatchery Mouth 0.57 54.75 93.53 Cabins near near Spring- Upper Slaughter Franklin burg field burg Sycamore effluent Bass Moore- Moorefield field Tract -house STP STP Gap near Plasma field Discharge Masonville Anhydroerythromycin WW WW WW Ciprofloxacin WW Erythromycin WW WW Lincomycin WW WW Ofloxacin WW WW WW WW Sulfamethoxazole WW WW Trimethoprim WW WW Tylosin WW WW Pentachloroanisole FT PS PS PS PS PS PS (PCA) 4-tert-octylphenol PL Diethoxynonylphenol WCest WCest Polybrominated diphenyl ether FT FT FT PS, PS PS, PS, PS, PS PL congeners (PBDEs) Arsenic FT FT WC WC WC WC WC Caffeine WCest WCest Chromium FT FT Mercury FT FT 1,3- FT dimethylnaphthalene 1,6- FT dimethylnaphthalene 1-methylnapthalene FT PSest. PS PS 2,6- FT PSest. PS PS dimethylnaphthalene

5/2/07 48 DRAFT: DO NOT CITE OR QUOTE Table A-1 cont.

North Fork Shenandoah (river South Branch Potomac miles) Spring South Pilgrim's At Run Small Fork Mainstem Near Near Pride Peters- Moore- Peters- At Hatchery Mouth 0.57 54.75 93.53 Cabins near near Spring- Upper Slaughter Franklin burg field burg Sycamore effluent Bass Moore- Moorefield field Tract -house STP STP Gap near Plasma field Discharge Masonville 2-methyl naphthalene FT PSest. PS PS PL Acenaphthene FT Fluoranthene FT est Naphthalene WC ,PS est. FT est PS PL PAH FT Phenanthrene FT PS PS PS Pyrene FT PS PS PS PCB FT FT FT MethTriclosan FT FT Tonalide (AHTN) PS PS PS Methyl salicylate WCest WCest WCest Chlordane FT FT FT PS PS PS PS PS PS Chlorpyrifos PS PS PS Nonachlor PS PS PS PS PS PS DDT and metabolites FT FT FT Fipronil gamma-HCH PS PS PS PS PS Hexachlorobenzene FT PS PS PS PS PS PS Trifluralin PS PS PS PS 1,4-Dichloro benzene WCest WCest WCest diethyl phthalate PL diethylhexyl phthalate PL Acetophenone PSest. PS, PS PL d-limonene PL Isophorone PS PS,WCest PS Tributyl phosphate WCest est 1= Concentration estimated; FT = Fish Tissue; PL = Smallmouth bass plasma; PS = Passive sampler ; WC = Water Column; WW = Waste water (effluent sample)

5/2/07 49 DRAFT: DO NOT CITE OR QUOTE 1 Discussion Although there was some evidence of As exposure in 2005, this was one 2 year before the 2006 fish kill. Ba was not observed in fish tissue. Intersex fish were 3 observed on the NF in 2005 which suggests that some chemical contaminant was 4 available at biologically active concentrations. No other data were available that would 5 constitute evidence. Neither pathology data from dead fish of the 2006 kill nor intersex 6 data were yet available. 7 8 Score NE, The available data provide no evidence of exposure or biological 9 mechanism. 10 11 A.1.8.3. Evidence from the Case: Complete Causal Pathway — 12 Data Used Locations of potential sources constitute evidence of potential causal 13 pathways. 14 National Pollution Discharge Elimination System (NPDES) discharges provide evidence 15 of sources of waste water (Figure A-8). 16 Agricultural sources were described in Bill van Wart’s presentation at the September 17 2006 CADDIS workshop, derived from the U.S. Census of Agriculture. 18 Industrial chemicals used in the watershed are indicated by the Toxics Release 19 Inventory. 20 21 Analysis NA. 22 23 Discussion Sources are summarized in Table A-2. 24 National Pollution Discharge Elimination System (NPDES) discharges constitute 25 evidence of potentially toxic chemicals. Although discharge permits do not allow know 26 toxic discharges, unknown discharge constituents or unreported exceedances of 27 permitted levels may result in toxic exposures. 28 Virginia did not have a comprehensive background document for the North Fork 29 Shenandoah River; however, Turner referred to the Bill Van Wart presentation. The 30 Shenandoah Basin, both North and South Forks, has a high concentration of livestock- 31 based agriculture, both poultry and cattle (van Wart presentation).

5/2/07 50 DRAFT: DO NOT CITE OR QUOTE 1

TABLE A-2

Potential Sources of Toxic Substances in North Fork Shenandoah

Source Potential Toxic Stressors Poultry farms, poultry processing plants Pesticides, antibiotics, cleaners Agriculture Pesticides, metals Highways PAHs, metals POTWS Pesticides, pharmaceuticals, PAHs, metals Industrial Unregulated substances, accidental discharges Legacy industrial PCB, other chlorinated organics, metals, PAHs, etc. Accidental/illegal dumping Any Golf courses Pesticides 2 3 4 The fish kills in the North Fork do not correspond to the location of POTWs or but other 5 discharges cannot be eliminated (Figure A-8). 6 7 Chicken processing houses may ship wastes via pretreatment plants to POTW. 8 Constituents of those wastes were unknown to workshop participants and should be 9 investigated. 10 11 Toxic Release Inventory (TRI) data suggest other possible water pollutants. See Tables 12 A-3, A-4 and A-5 for 2005 releases. 13 14 Score + Evidence somewhat supports because the identified sources constitute at least 15 one step in each causal pathway (Figure 5).

5/2/07 51 DRAFT: DO NOT CITE OR QUOTE 1 TABLE A-3

2005 TRI-eFDR Reported Releases to Streams, POTW, or Other for Facilities Within the Two Watersheds

Quantity County Facility Name Compound(s) Disposal Method (pounds) Shenandoah GLOBAL STONE LEAD 4 Other onsite CHEMSTONE CORP COMPOUNDS (Air/Stormwater also) PERRY JUDD'S INC. CERTAIN GLYCOL STRASBURG WATER STRASBURG DIV ETHERS 200 TREATMENT PLANT (also Air emissions) ETHYLENE STRASBURG WATER GLYCOL 6,100 TREATMENT PLANT (Also Air emissions) GEORGE'S NITRATE 123,828 Stoney Creek CHICKEN LLC COMPOUNDS (also Offsite) Pendleton GREER LEAD 2,167 other onsite INDUSTRIES INC. COMPOUNDS (Air / Stormwater also) DBA GREER LIME CO MERCURY 16 other onsite COMPOUNDS (Air / Stormwater also) Hardy PILGRIM'S PRIDE AMMONIA South Fork of the South 171 CORP MOORFIELD Branch of Potomac PERPARED FOOD (Air / Offsite also) PLANT NITRATE South Fork of the South 93,583 COMPOUNDS Branch of Potomac (Offsite also) PILGRIM'S PRIDE NITRATE South Fork of the South 127,000 CORP COMPOUNDS Branch of Potomac MOOREFIELD FRESH FACILITY (Offsite also) 2

5/2/07 52 DRAFT: DO NOT CITE OR QUOTE 1 TABLE A-4

2005 TRI-eFDR Reported Releases for Dominion Mount Storm Power Station, Grant County

Compound Quantity (pounds) Disposal Ammonia 1,200 Stony River 36,000 onsite landfill (also air emissions and offiste) Arsenic Compounds 150 Stony River (also air, landfill, offsite) Barium Compounds 630,000 onsite landfill (also air, offsite) Beryllium Compounds 9,200 onsite landfill (air, offsite also) Chromium Compounds 110,000 onsite landfill (except chromite ore mined in (also air, offsite) the Transvaal Region) Cobalt Compounds 40,000 onsite landfill (also air, offsite) Copper Compounds 11 Stony River 130,000 onsite landfill (also air, offsite) Lead Compounds 1 Stony River 61,662 onsite landfill (air, offsite) Manganese Compounds 1,200 Stony River 160,000 onsite landfill (air, offsite) Mercury Compounds 5 Stony River 1,358 onsite landfill (air, offsite) Nickel Compounds 1 Stony River 115,000 onsite landfill (air, offsite) Selenium Compounds 170 Stony River 170,000 onsite landfill (air, offsite) Vanadium Compounds 240,000 onsite landfill (offsite) Zinc Compounds 160 Stony River 140,000 onsite landfill (air, offsite)

5/2/07 53 DRAFT: DO NOT CITE OR QUOTE 1 TABLE A-5

Other Compounds Found in 2005 TRI-eFDR for the Two Watersheds

These compounds were disposed of either in Air, Offsite, or reported on Form A and thus have no disposal information: 1,2,4-Trimethylbenzene Antimony Compounds Benzene Chlorine Copper Diisocyanates Dioxin and Dioxin-like Compounds Ethylbenzene Hydrochloric Acid Hydrogen Fluoride Lead Methyl Isobutyl Ketone Mercury Methanol Methyl Tert-Butyl Ether Mixture n-Butyl Alcohol n-Hexane Naphthalene Ozone Polycyclic Aromatic Compounds Styrene Sulfuric Acid Toluene Xylene (Mixed Isomers)

5/2/07 54 DRAFT: DO NOT CITE OR QUOTE 1 2 FIGURE A-8 3 Locations of All Permitted Waste-Water Discharges on the North Fork 4 Shenandoah River

5/2/07 55 DRAFT: DO NOT CITE OR QUOTE 1 A.1.8.4. Evidence from the Case: Tests of Media in Laboratory — 2 Data Used U.S. EPA toxicity tests of South Branch media (Amy Bergdale). 3 4 Analysis None. 5 6 Discussion Toxicity tests were performed on media from the North Fork Shenandoah 7 and the Cowpasture River, a reference site. Both Ceriodaphnia and fathead minnows 8 were tested. Results were equivocal, with some observed toxicity but toxicity was 9 similar in the reference river as in the North Fork of the Shenandoah. The samples for 10 the toxicity tests were taken after the fish kills. 11 12 Score NE Toxicity tests were on post-kill waters. 13 14 A.1.8.6. Evidence from other Studies: Stressor-Response from Laboratory 15 Studies — 16 Data Used Although data are available concerning the toxicity of many potential 17 contaminants, ambient concentrations are not available for comparison. 18 19 Analysis NA. 20 21 Discussion NA. 22 23 Score NE, No evidence. 24 25 A.1.8.7. Evidence from Other Studies: Analogous Cases — 26 Data Used None 27 28 Analysis NA. 29 30 Discussion Prior kills in the Shenandoah could be analyzed as analogous cases, but 31 data for chemical exposures from those kills are lacking.

5/2/07 56 DRAFT: DO NOT CITE OR QUOTE 1 2 Score NE. 3 4 A.1.9. Candidate Cause: Starvation. 5 A.1.9.1. Evidence from the Case: Spatial/Temporal Co-occurrence — 6 Data Used Length, weight and age observations of smallmouth bass before, during and 7 after kill. 8 9 Analysis NA. 10 11 Discussion Very low weight relative to length is considered the measure of starvation, 12 not just a symptom. Observations on weights of dead smallmouth bass showed no 13 indication of low weight or weight loss compared to before the kill, after the kill, or in 14 other rivers. Killed fish did not appear different from healthy fish in terms of body 15 weight. 16 17 Score - - - Evidence strongly suggests that the candidate cause did not occur. 18 19 A.1.9.2. Evidence from the Case: Evidence of Exposure or Biological 20 Mechanisms — 21 NE, No evidence exists on stomach contents of fish in the 2006 fish kills. 22 23 A.1.9.3. Evidence from the Case: Complete Causal Pathway — 24 NE, No evidence exists on food availability in the North Fork of the Shenandoah. 25 26 A.1.9.4. Evidence from Other Studies: Analogous Cases — 27 Data Used Unpublished 2005 letter report to Don Kain of the Virginia DEQ, Valley 28 Regional Office, from Dr. Stephen Smith of Virginia Polytechnic Institute and State 29 University (Virginia Tech). 30 31 Analysis NA.

5/2/07 57 DRAFT: DO NOT CITE OR QUOTE 1 Discussion In the 2005 South Fork Shenandoah fish kills, dead long-eared sunfish, 2 smallmouth bass and suckers all had food in their stomachs, appeared normal in body 3 weight and had normal body fat. 4 5 Score Negative, -. Candidate cause did not occur in an analogous kill. 6 7 A.2. CASE 2—2006 SOUTH BRANCH POTOMAC RIVER, WV 8 A.2.1. Anoxia due to Low Dissolved Oxygen. 9 A.2.1.1. Spatial/Temporal Co-occurrence — 10 Data Used USGS continuous YSI D.O. monitor spatial and temporal data from the fish 11 mortality event; temporal data from a reference control site on the South Fork 12 Shenandoah River. (Data on team room) Ad hoc measurements during the kill. 13 14 Analysis Co-occurrence inferred from observations of dying fish and concurrent ad hoc 15 DO measurements of 8 mg/L by Jim Hendrick. 16 17 Discussion The co-occurrence of dying fish and normal dissolved oxygen levels refutes 18 the possibility that low dissolved oxygen caused anoxia. Mortality from low dissolved 19 oxygen is rapid. Therefore, if DO concentrations are high but fish are still dying, low DO 20 cannot be the cause. 21 22 Score R, The evidence is sufficient to refute low dissolved oxygen concentrations as 23 the cause. 24 25 A.2.2. Anoxia due to Gill Injury (Gill Hyperplasia). 26 A.2.2.1. Spatial/Temporal Co-occurrence — 27 Data Used Gross and histopathology samples and data from the mortality event and 28 from reference sites on the Greenbrier River were taken. However, those data were not 29 available for this analysis. 30 31 Analysis Pending pathology data.

5/2/07 58 DRAFT: DO NOT CITE OR QUOTE 1 Discussion None. 2 3 Score NE (P), No Evidence at this time. 4 5 A.2.2.2. Evidence of Exposure or Biological Mechanism — 6 Data Used Observational data from Jim Hendrick (WV DNR) on affected fish 7 demonstrating clinical signs of gasping, abnormal surfacing, lethargy and decreased 8 flight behavior response. 9 10 Analysis Minimal analysis due to non-specific observational nature of the data. 11 12 Discussion Some fish displayed gasping behavior but most were simply lethargic. 13 14 Score + Data are positive but weak and inconsistent. 15 16 A.2.2.3. Causal Pathway — 17 Data Used Data was obtained on water pH (including pH fluctuations by USGS), 18 changes in water temperature, water ammonia levels, water nitrite / nitrate levels, total 19 suspended solids (WV Dept of Agriculture), possible parasite data (unavailable 20 pathology reports). 21 22 Analysis Pending. 23 24 Discussion Gill injury could have been caused by several causal pathways. Evidence 25 for occurrence of a causal pathway consists of the occurrence of an agent in a pathway 26 leading to the proximate cause. Some of the agents in the pathway to this candidate

27 cause are also candidate causes themselves (i.e., pH, NH3, and pathogens). However, 28 the evidential requirements here are different. Elevated concentrations of an agent that 29 is a proximate cause must occur at the time of a kill, but agents in a causal pathway 30 may have been elevated in the past. It is necessary only that their effects persist.

5/2/07 59 DRAFT: DO NOT CITE OR QUOTE 1 pH – Data showed wide pH fluctuations on a daily basis, pH levels went to mid 9's. 2 Fluctuations were high immediately before the kill but were even higher at the 3 beginning of April (Section A.2.6). pH fluctuations may have contributed to gill injury 4 and stress to the fish through alkalosis or ammonia autointoxication.

5 NH3 – Ammonia was not elevated at the time of or immediately before the kill. Toxic gill 6 necrosis can result from autointoxication (metabolism of proteins in excess of 7 excretory capacity) which may be triggered by starvation, a sudden drop in 8 temperature, an increase in pH above 9, an increase in aqueous ammonia, or an 9 increase in protein in the diet (Smutna et al., 2002). Of these, only elevated pH was 10 observed (see above). 11 Nitrate and Nitrite – Data were unavailable, but may be elevated by nitrification of 12 ammonia in the watershed. 13 Temperature – Water temperature rose, which increases biological activity, including 14 oxygen requirements and most toxic responses. 15 Auto-intoxication – The elevated pH and temperature but not ammonia suggest that 16 conditions provide ambiguous evidence for an ammonia auto intoxication cycle to be 17 initiated internally in the fish. 18 Conductance – Conductance was not elevated. 19 Pathogens – Data were also available from analyses of bacterial and protist 20 communities in mucus from fish with and without lesions collected during the 21 mortality event (Gillevet et al., 2006). Parasite levels were potentially increased, 22 also adding further physiological stressor, but pathology data were unavailable. 23 Stress – The kill did not co-occur with spawning and seasonal behavior was normal. 24 Flow – A high flow event occurred prior to the kill, which could carry contaminants into 25 the river or resuspend material in the river potentially injuring the gills. 26 Abrasion – Concentrations of total suspended solids were not high. Neither periphytic 27 nor planktonic diatoms appeared to be heavy. 28 29 Score Water pH 0 30 Water Ammonia 0 31 Water Nitrate & Nitrite 0 32 Water Temperature +

5/2/07 60 DRAFT: DO NOT CITE OR QUOTE 1 Auto intoxication cycle 0 2 Conductance - 3 Stress, Behavioral - 4 Abrasion, Flow event + 5 Abrasion, Total Suspended Solids - 6 Abrasion, Planktonic diatoms - 7 Abrasion, Wash off of dead diatoms - 8 9 A.2.2.4. Stressor-Response from Laboratory Studies — 10 Data Used Toxicological benchmarks from laboratory tests of fish were obtained for pH,

11 NH3, and temperature. 12 13 pH 14 9 causes carp mortality (Schaperclause, 1952) 15 When water pH > blood pH, NH3 excretion is reduced; above pH 9.5 it is blocked 16 (Schaperclause, 1952; Wood, 2001). 17 Other effects of high pH at the gills include increased CO2 excretion leading to 18 alkalosis and reduced Na+ and Cl- uptake (Wood, 2001). 19 Reported gill injuries were limited to hyperplasia of the chloride cells, not gross 20 hyperplasia (Wood, 2001). 21 22 Temperature 23 Rapid changes in temperature can increase morbidity and mortality in fish 24 (Stoskopf, 1993). 25

26 NH3 27 0.2 mg/L caused gill injury in Brown trout (U.S. EPA, 1999). 28 29 Gill lamellae obtained from parental fish exposed to un-ionized ammonia

30 concentrations ranging from 0.02 mg NH3-N/L to 0.05 mg NH3-N/L for four months, and

31 0.05 mg NH3-N/L and 0.06 mg NH3-N/L for seven and eleven months, showed mild to 32 moderate fusion, aneurysms, and separation of the epithelia from the underlying 33 basement membrane (U.S. EPA, 1999). 34 “In contrast to acute exposures, a variety of morphologic changes in the gills 35 have been described during chronic sublethal ammonia exposure. Most prominent are 36 an overall swelling of the respiratory lamellae, proliferation of epithelial cells, increased 37 diffusion distance, and an increased prevalence of bacterial gill disease … These

5/2/07 61 DRAFT: DO NOT CITE OR QUOTE 1 responses would be expected to decrease the respiratory gas exchange capacity of the 2 fish, and thus its swimming performance and tolerance to hypoxia” (Wood, 2001). 3 Smallmouth bass are fairly sensitive. The growth effect concentrations ranged

4 from 0.05 mg NH3-N/L at pH=6.6 to 0.71 mg NH3-N/L at pH=8.68 (U.S. EPA, 1999, pg. 5 118). 6 Red horse sensitivities are unknown. 7 As fish emerge from torpor in the spring, toxic ammonia concentrations drop from 8 1.7 to 0.2 mg/L (Schaperclause, 1952). 9 0.5 mg/L total ammonia highest observed in spring. 10 11 Nitrite and Nitrate 12 Chronic exposures to nitrite can cause gill hyperplasia (Kroupova et al., 2005). 13 14 Analysis Comparison of aqueous concentrations of stressors to concentrations causing 15 effects in laboratory studies. 16 17 Discussion Data were not sufficiently available at the workshop to confidently evaluate 18 the potential for observed levels of stressors, singly or in combination, to injure gills. 19 Although the type of injury is unknown and the laboratory studies are not clearly 20 comparable to the field, the available evidence suggests that chronic exposures to 21 some contaminants may be sufficient. Also differences in sensitivity in the laboratory 22 should be compared to apparent differences in response among species during the kill. 23 24 Score +(P) Some evidence supports the candidate cause but more is Pending. 25 26 A.2.2.5. Analogous Cases — 27 Data Used NA. 28 29 Analysis NA. 30

5/2/07 62 DRAFT: DO NOT CITE OR QUOTE 1 Discussion A kill in the South Branch in 2002 primarily affecting red horse, but also 2 including smallmouth bass, and redbreast sunfish occurred in the same time frame as 3 the 2006 kill. Also like the 2006 kill, it also occurred after a spring run-off event. Flow, 4 pH & temp data maybe available for the 2002 kill. 5 6 Score NE(P) The 2002 kill may be analogous but data were not available to evaluate 7 the candidate cause. 8 9 A.2.3. Anoxia due to Low Blood Oxygen Affinity (Methemoglobinemia). 10 A.2.3.1. Spatial/Temporal Co-occurrence — 11 Data Used Plasma samples archived and observational data were taken from the fish 12 mortality event and reference sites on the Greenbrier River, but those data were 13 unavailable. 14 Observations by State biologists were recounted at the workshop. 15 16 Analysis None. 17 18 Discussion Neither brown gills nor brown blood were not observed in these fish by 19 State biologists. High pH can also cause low blood affinity without brown blood. 20 Pathology reports could provide more definitive data. 21 22 Score -/NE (P) Field observations were negative but more definitive data from the 23 pathology reports are pending. 24 25 A.2.3.2. Evidence of Exposure or Biological Mechanism — 26 Data Used Observational data from Jim Hendrick (WV DNR) on affected fish 27 demonstrating clinical signs of gasping, abnormal surfacing, lethargy and decreased 28 flight behavior response. 29 30 Analysis Minimal analysis due to non-specific observational nature of data. 31

5/2/07 63 DRAFT: DO NOT CITE OR QUOTE 1 Discussion Some fish displayed gasping behavior but most were simply lethargic. 2 3 Score + Data somewhat support methemoglobinemia but are inconsistent. 4 5 A.2.3.3. Causal Pathway — 6 Data Used Knowledge of sources and processes in the watershed. 7 8 Analysis None. 9 10 Discussion Nitrite accumulates in water when ammonia concentrations are elevated 11 and the second stage of nitrification is inhibited. Sources of ammonia are present in the 12 watershed, but nitrification rates or processes controlling the rates are unknown. 13 14 Score 0, The evidence is ambiguous because sources are known but not 15 transformation processes. 16 17 A.2.3.4. Stressor-Response from Laboratory Studies — 18 Data Used Literature reviews 19 Methemoglobinemia symptoms occur at 0.10 to 0.50 mg/L in sensitive species

20 (channel catfish and trout) and LC50 values range from 0.60 to 200 mg/L (Animal 21 Disease Diagnostic Laboratory, 1998). 22 Centrarchids (includes sunfish and black bass) are refractory to 23 methemoglobinemia (Kahn, 2005). 24 25 Analysis Logic and comparison to ambient concentrations. 26 27 Discussion The maximum reported nitrite concentrations in the South Branch during 28 the period of the kill was 0.0041 mg/L. That is less than half the concentrations 29 reported to cause methemoglobinemia in laboratory studies and less the a tenth of 30 concentrations that cause lethal methemoglobinemia in sensitive species.

5/2/07 64 DRAFT: DO NOT CITE OR QUOTE 1 This kill involved catastomids rather than centrarchids so the relative sensitivities may 2 be consistent. 3 4 Score - - This evidence significantly weakens methemoglobinemia as a candidate 5 cause. Reports of blood color in pathology reports would be definitive. 6 7 A.2.4. Mortality due to Other Pathogenic Modes of Action. 8 A.2.4.1. Spatial/Temporal Co-occurrence — 9 Data Used Gross and histopathology samples and data from mortality event and from 10 reference sites on the Greenbrier River. However, those data were not available for this 11 analysis. 12 13 Analysis None, pending pathology data. 14 15 Discussion None. 16 17 Score NE (P) No Evidence, but data are pending. 18 19 A.2.4.2. Evidence of Exposure or Biological Mechanism — 20 Data Used Observational data on affected fish demonstrating lethargy and decreased 21 flight behavior response. Pathology data were unavailable. 22 23 Analysis Minimal analysis due to non-specific observational nature of data. 24 25 Discussion Observational data is in this case is diagnostically non-specific. 26 27 Score 0/NE(P) Ambiguous behavior, other data are pending. 28 29 A.2.4.3. Complete Exposure Pathway — 30 Data Used Knowledge of State biologists. 31

5/2/07 65 DRAFT: DO NOT CITE OR QUOTE 1 Analysis None. 2 3 Discussion Sources of pathogens may include stocked fish, released bait fish or 4 effluents from hatcheries. Trout but not smallmouth bass are stocked in the South 5 Branch. Bait minnows inevitably are released. Hatcheries occur at and below the kill 6 site. Spring Run Hatchery is located upstream of Petersburg on South Fork of Mill Run 7 of South Branch. 8 9 Score + Some steps are present. 10 11 A.2.5. Mortality due to High pH. 12 A.2.5.1. Spatial/Temporal Co-occurrence Evidence (Specifically Temporal Co- 13 occurrence) — 14 Data Used Continuously monitored pH data from January-July 2006 on the South 15 Branch of the Potomac at Springfield (mile 6) collected by the USGS. 16 17 Analysis Continuously monitored pH data plotted vs. date and with respect to the kill 18 interval (Figure A-9). 19 South Branch Potomac, Springfield 20 10.0 [2006 data]

21 m mu

i 9.0

22 x

23 ma H

p 8.0 24 ily a 25 D 7.0 26 12/19 2/7 3/29 5/18 7/7 27 Date 28 29 FIGURE A-9 30 A Plot of the Daily Maximum pH Balues in the South Branch Potomac River at 31 Springfield. The shaded band covers the period of the kill. 32

5/2/07 66 DRAFT: DO NOT CITE OR QUOTE 1 Discussion pH values were high (above 9) beginning as early as January, however 2 there was no kill at this time (Figure A-9). The kill did not occur until late May when pH 3 values were actually dropping below the March highs. All of this demonstrates a lack of 4 temporal alignment of peaks in pH and the onset of the fish kill in the South Branch. 5 6 Score --- The effect both does not occur when the candidate cause occurs, and does 7 occur when the candidate cause does not occur. 8 9 A.2.5.2. Stressor-Response Relationship in the Lab — 10 Data Used Scott, Lucas, and Wilson in Aquatic Toxicology, 2005 addressed effects of 11 pH as high as 9.5 on behavior of rainbow trout and perch. 12 13 Analysis Maximum pH values during the case of interest were compared with those 14 used during this study. 15 16 Discussion The pH maximum during the reported experiment was as high as 9.5, 17 similar to what we’ve observed on the South Branch in 2006 during the kill. Since fish 18 exposed to that pH were shown to have reduced ability to excrete ammonia (as 19 measured in the fishes’ blood), but exhibited no lethality or change in swimming 20 behavior, we have to conclude that our fish should not have been dying due to 21 comparatively lower pH levels alone. Note that this experiment was carried out on 22 different species. 23 24 Score - Little effect gradient is observed relative to exposure to the candidate cause, in 25 a controlled lab environment. The fish exposed to pH levels similar to those found in 26 our case of interest did showed reduced ammonia excretion capabilities, but did not die. 27 28 A.2.5.3. Stressor-Response Relationship in the Lab — 29 Data Used An article by Serafy and Harrell (1993) addressing sublethal effects of high 30 pH on bluegill, striped bass, and killifish. 31

5/2/07 67 DRAFT: DO NOT CITE OR QUOTE 1 Analysis Comparison of fish behavior in our case of interest with that observed in the 2 experiment under similar pH conditions. 3 4 Discussion The fish in this experiment were subjected to pH increases of about 1 unit 5 over the course of less than one hour, with the highest replicate reaching a final pH of 6 about 9.5. Because the exposures are so short in duration, the comparability with our 7 case is severely reduced. Also, while these results may be useful for assessing fish 8 stress response, they are not very useful for assessing lethality since none of the fish 9 died. 10 11 Score - The experiment shows that pHs as high as 9.5, similar to the peaks measured 12 in our case of interest, are not lethal to fish. However, the key differences of conditions 13 in each scenario reduce the comparability of each situation to the other. 14 15 A.2.6. Mortality due to pH Fluctuations. 16 A.2.6.1. Spatial/Temporal Co-occurrence (Specifically Temporal Co- 17 occurrence) — 18 Data Used Continuously monitored pH data from January-July 2006 on the South 19 Branch of the Potomac at Springfield (mile 6) collected by the USGS. 20 21 Analysis Calculated pH shifts over 24 hour windows plotted against the date (Figure 22 A-10). 23 24 Discussion pH fluctuations were high (0.2-1.2 units) before the kill, dropping somewhat 25 after the kill, and with the highest fluctuations coming around early April. Though there 26 may be some interaction/combination of temperature and pH effects on the fish, when 27 considering pH fluctuations alone, there is not significant temporal co-occurrence 28 evidence to support pH fluctuations as a candidate cause. 29 30 Score --- The effect both does not occur when the candidate cause occurs, and does 31 occur when the candidate cause does not occur.

5/2/07 68 DRAFT: DO NOT CITE OR QUOTE 1 South Branch Potomac, Springfield 1.2 2 [2006 data]

3 n io t a u

4 t c u l 0.6

5 f H

6 p ily a

7 D 8 0.0 12/19 2/7 3/29 5/18 7/7 9 Date 10 11 FIGURE A-10 12 A plot of the differences between daily maximum and minimum pH values in the South 13 Branch Potomac River at Springfield. The shaded band covers the period of the kill. 14 15 16 A.2.6.2. Stressor-Response Relationship in the Field — 17 Data Used An article by Serafy and Harrell (1993) addressing sublethal effects of pH 18 fluctuations on natural fish communities including pumpkinseed, eel, and killifish. 19 20 Analysis Comparison of fish behavior in our case of interest with that observed in the 21 experiment under similar pH conditions. 22 23 Discussion The fish communities in this experiment were monitored during a natural 24 shift in pH through the course of a day while pH was monitored. The observers noted 25 shifts in community density, biomass and species richness. None of these changed 26 drastically through the course of the pH shift. pH was measured as varying from ~8.5 27 up to a maximum of ~9.8 and then down to a minimum of ~7.6. The pH regime in this 28 experiment is probably the most relevant external scenario to ours because of the 29 magnitude of pH shift and the high average pH around which the pH was centered. On 30 the other hand, the other environmental/ecological factors in this setup decreased 31 comparability between the experimental scenario and ours on the South Branch. The 32 study was conducted in a tidal freshwater environment of the Chesapeake Bay with

5/2/07 69 DRAFT: DO NOT CITE OR QUOTE 1 different fish species. We must question whether, if this pH regime is natural and 2 common for that area, then are the native fish more accustomed and better adapted to 3 these conditions? Also, the observers noted whether or not fish fled the study site, 4 which was a hydrilla grass bed, and interpreted this as a potential reaction to 5 undesirable pH levels. But if pH remained the same at farther distances from the grass 6 bed, we shouldn’t expect fleeing the grass bed to be of any benefit, so perhaps fleeing 7 is an inappropriate endpoint of assessment. Also, there could have been parameters 8 other than pH that weren’t measured that influenced the fishes’ decision to stay or 9 leave. 10 11 Score - The experiment shows that pHs fluctuations of over 2 units centered around a 12 pH of ~8.5, similar to the fluctuations measured in our case of interest, are not lethal to 13 fish. However, the key differences of conditions in each scenario reduce the 14 comparability of each situation to the other. 15 16 A.2.6.3. Stressor-Response Relationship in the Field — 17 Data Used Continuously monitored pH readings from USGS gage # 1668000 on the 18 Rappahannock River from January 2006 to October 2006. 19 20 Analysis Calculated daily pH shift plotted against time (Figure A-11). 21 22 Discussion The Rappahannock River is being used here as an out-of-basin reference 23 stream, though it should be noted that there is no solid evidence of similarities in 24 geology, ecology, and water quality between this reference stream and the South 25 Branch. That being said, this data shows pH fluctuations of even greater magnitude 26 (1.5-2 units) than those seen on the South Branch and there were no reported kills 27 there. 28

5/2/07 70 DRAFT: DO NOT CITE OR QUOTE 1 2 3.0 Rappahannock River [2006 USGS data (provisional)] n

3 io t a

u 2.0 4 t c u l

5 f H

p 1.0 6 ily a 7 D 0.0 8 1/3 2/22 4/13 6/2 7/22 9/10 10/30 9 Date 10 11 FIGURE A-11 12 A Plot of the Differences Between Daily Maximum and Minimum pH Values in the 13 Rappahannock River, a Reference Stream. The shaded band covers the period of the 14 kill in the South Branch. 15 16 17 Score - A differential in response relative to exposure to the candidate cause was 18 observed at non-spatially linked sites, but the difference is not in the expected direction. 19 The Rappahannock is a non-spatially linked site and shows greater pH fluctuation than 20 the South Branch, but with no increase in fish kill occurrence (in fact, no occurrence of 21 fish kills). This was not given a stronger negative score because of the weak similarities 22 in conditions at each of the two sites, and therefore the low suitability of the 23 Rappahannock as a reference site. 24 25 A.2.7. Mortality due to High Ammonia Concentrations. 26 A.2.7.1. Spatial/Temporal Co-occurrence (Specifically Temporal Co- 27 occurrence) — 28 Data Used Ammonia sampled 5 times monthly by the West Virginia Department of 29 Agriculture at random times of the month from October 2005-October 2006 at 11 sites 30 on the South Branch of the Potomac River including: SB11 Petersburg, SB13 Potomac 31 Valley View, SB15 below Moorefield, SB16 Old Fields Bridge, SB17 McNeil, SB18

5/2/07 71 DRAFT: DO NOT CITE OR QUOTE 1 Harrisons, SB19 Stony Run, SB20 Romney Bridge, SB22 Blues Beach Bridge, SB24 2 Blue Ford North, and SB26 at the mouth (Figure A-12). 3

4 5 FIGURE A-12 6 Locations of Ammonia Sampling Stations in the South Branch of the Potomac River 7 8 9 Analysis Unionized ammonia concentrations were calculated from the total ammonia 10 measurements using temperature and pH measurements taken at the time of sampling. 11 Both unionized and total ammonia concentrations were plotted against time (Figure A- 12 13 through A-23).

5/2/07 72 DRAFT: DO NOT CITE OR QUOTE 1 1.0E+00 South Branch Potomac, below Moorefield (SB15)

2 L) [2005-2006 WV Dept. of Ag. Data] g/ m

3 ( 1.0E-02 a ni

4 o

m 1.0E-04 5 m d a e

6 z 1.0E-06 oni 7 i n 8 U 1.0E-08 8/6 11/14 2/22 6/2 9/10 12/19 9 Date 10 11 FIGURE A-13 12 A Plot of Unionized Ammonia Concentrations in the South Branch Potomac River Below 13 Moorefield. The shaded band covers the period of the kill. 14 15 1.0E+00 South Branch Potomac, Old Fields Bridge (SB16) 16 L) [2005-2006 WV Dept. of Ag. Data] g/

17 m

( 1.0E-02 a

18 ni o

19 m 1.0E-04 m d a

20 e

z 1.0E-06

21 oni i n

22 U 1.0E-08 8/6 11/14 2/22 6/2 9/10 12/19 23 Date 24 25 FIGURE A-14 26 A Plot of Unionized Ammonia Concentrations in the South Branch Potomac River at Old 27 Fields Bridge. The shaded band covers the period of the kill. 28

5/2/07 73 DRAFT: DO NOT CITE OR QUOTE 1 1.0E+00 South Branch Potomac, McNeil (SB17)

2 L) [2005-2006 WV Dept. of Ag. Data] g/

3 m

( 1.0E-02 a

4 ni o

m 1.0E-04

5 m d a

6 e

z 1.0E-06 oni

7 i n 8 U 1.0E-08 8/6 11/14 2/22 6/2 9/10 12/19 9 Date 10 FIGURE A-15 11 A Plot of Unionized Ammonia Concentrations in the South Branch Potomac River at 12 McNeill. The shaded band covers the period of the kill. 13 14 15 16 1.0E+00 South Branch Potomac, Harrisons (SB18)

17 L) [2005-2006 WV Dept. of Ag. Data] g/ m

18 ( 1.0E-02 a ni

19 o

m 1.0E-04 20 m d a e

21 z 1.0E-06 oni 22 i n 23 U 1.0E-08 8/6 11/14 2/22 6/2 9/10 12/19 24 Date 25 FIGURE A-16 26 A Plot of Unionized Ammonia Concentrations in the South Branch Potomac River at 27 Harrisons. The shaded band covers the period of the kill. 28

5/2/07 74 DRAFT: DO NOT CITE OR QUOTE 1 2 1.0E+00 South Branch Potomac, Stony Run (SB19)

L) [2005-2006 WV Dept. of Ag. Data] 3 g/ m

( 1.0E-02 a

4 ni o

5 m 1.0E-04 m

6 d a e

z 1.0E-06 7 oni i n

8 U 1.0E-08 9 8/6 11/14 2/22 6/2 9/10 12/19 Date 10 11 FIGURE A-17 12 A Plot of Unionized Ammonia Concentrations in the South Branch Potomac River at 13 Stony Run. The shaded band covers the period of the kill. 14 15 16 17 1.0E+00 South Branch Potomac, Romney Bridge (SB20)

18 L) [2005-2006 WV Dept. of Ag. Data] g/

19 m

( 1.0E-02 a

20 ni o

m 1.0E-04

21 m d a

22 e

z 1.0E-06

23 oni i n 24 U 1.0E-08 8/6 11/14 2/22 6/2 9/10 12/19 25 Date 26 FIGURE A-18 27 A Plot of Unionized Ammonia Concentrations in the South Branch Potomac River at 28 Romney Bridge. The shaded band covers the period of the kill. 29

5/2/07 75 DRAFT: DO NOT CITE OR QUOTE 1 1.0E+00 South Branch Potomac, Blues Beach Bridge (SB22)

2 L) [2005-2006 WV Dept. of Ag. Data] g/ m

3 ( 1.0E-02 a ni

4 o m 1.0E-04 5 m d a e

6 z 1.0E-06 oni i

7 n

U 1.0E-08 8 8/6 11/14 2/22 6/2 9/10 12/19 9 Date 10 FIGURE A-19 11 A Plot of Unionized Ammonia Concentrations in the South Branch Potomac River at 12 Blues Beach Bridge. The shaded band covers the period of the kill. 13 14 15 16

17 1.0E+00 South Branch Potomac, Blue Ford North (SB24)

L) [2005-2006 WV Dept. of Ag. Data]

18 g/ m

( 1.0E-02

19 a ni 20 o m 1.0E-04 21 m d a e 22 z 1.0E-06 oni i

23 n U 1.0E-08 24 8/6 11/14 2/22 6/2 9/10 12/19 25 Date 26 FIGURE A-20 27 A Plot of Unionized Ammonia Concentrations in the South Branch Potomac River at 28 Blue Ford North. The shaded band covers the period of the kill. 29

5/2/07 76 DRAFT: DO NOT CITE OR QUOTE 1 2 1.0E+00 South Branch Potomac, mouth (SB26)

3 L) [2005-2006 WV Dept. of Ag. Data] g/ m

4 ( 1.0E-02 a ni

5 o

m 1.0E-04 6 m d a e

7 z 1.0E-06 oni 8 i n

U 1.0E-08 9 8/6 11/14 2/22 6/2 9/10 12/19 10 Date 11 FIGURE A-21 12 A Plot of Unionized Ammonia Concentrations in the South Branch Potomac River at its 13 Mouth. The shaded band covers the period of the kill. 14 15 16 South Branch Potomac, Petersburg (SB11) 17 1.0E+00 L) [2005-2006 WV Dept. of Ag. Data] g/

18 m

( 1.0E-02 a

19 ni o

20 m 1.0E-04 m

21 d a e

z 1.0E-06 22 oni i n

23 U 1.0E-08 24 8/6 11/14 2/22 6/2 9/10 12/19 Date 25 26 FIGURE A-22 27 A Plot of Unionized Ammonia Concentrations in the South Branch Potomac River at 28 Petersburg. The shaded band covers the period of the kill. 29

5/2/07 77 DRAFT: DO NOT CITE OR QUOTE 1 1.0E+00 South Branch Potomac, Potomac Valley View (SB13)

2 L) [2005-2006 WV Dept. of Ag. Data] g/

3 m

( 1.0E-02 a

4 ni o

m 1.0E-04

5 m d a

6 e

z 1.0E-06

7 oni i n 8 U 1.0E-08 8/6 11/14 2/22 6/2 9/10 12/19 9 Date 10 11 FIGURE A-23 12 A Plot of Unionized Ammonia Concentrations in the South Branch Potomac River at 13 Potomac Valley View. The shaded band covers the period of the kill. 14 15 16 Discussion Unionized ammonia values were calculated since this is the most toxic 17 form. The amount of total ammonia present as unionized ammonia increases with pH.

18 Also, since pH fluctuates through the day, if NH3 is sampled in the morning and pH is 19 recorded then, we should expect pH to rise by the afternoon, with a concurrent rise in

20 unionized NH3 concentration. Therefore, we may not be recording maximum unionized

21 NH3 values. However, macrophytes, periphyton and phytoplankton take up ammonia 22 during the day, and that process should also be considered. Hence, modeling of 23 maximum unionized ammonia concentrations would require modeling rates of uptake as 24 well as pH increases, which is not possible with available data and is beyond the scope 25 of this workshop. Still, data taken at all sites within the kill region show the same trend 26 of common unionized ammonia levels below 1µg/L with the highest levels often 27 between 1-10 µg/L (Figure A-24). The highest level recorded in the kill zone was 170 28 µg/L of unionized ammonia observed at site SB 15 just below Moorefield, but this 29 measurement was taken in August 2006. All sites in the kill zone had values below 5 30 µg/L during the kill. If unionized ammonia alone were responsible for the kill, we would 31 expect to see the peak of ammonia concentrations corresponding temporally with the 32 kill, which they do not.

5/2/07 78 DRAFT: DO NOT CITE OR QUOTE 1 Score --- The effect both does not occur when the candidate cause occurs, and does 2 occur when the candidate cause does not occur. 3 4 A.2.7.2. Stressor-Response Relationship in the Field — 5 Data Used Frequent measurements of ammonia at sites upstream of the kill on the 6 South Branch from October 2005 - October 2006. Rivers/sites sampled include the SB 7 11, at Petersburg, and SB 13 at Potomac Valley View. 8 9 Analysis Unionized ammonia levels are presented, calculated using pH and 10 temperature data. These values were plotted against time. 11 12 Discussion While the data do show slightly lower levels of ammonia upstream of the kill 13 zone (common levels between 0.005 µg/L and 0.5 µg/L) than those observed at 14 Moorefield and below (common levels of 0.05 µg/L to 5 µg/L), if we look at the date of 15 the kill specifically, we see that unionized ammonia levels upstream of the kill are within 16 1 order of magnitude of the levels observed at some sites within the kill zone. Also, 17 when upstream sites’ ammonia levels peak in July/August, they approach 5 µg/L, which 18 is higher than some sites’ levels within the kill zone during the kill, yet we see no kills 19 upstream of Moorefield. 20 21 Score - Little effect gradient is observed relative to exposure to the candidate cause. 22 The upstream reference sites here show similar ammonia concentrations to those 23 downstream in the kill zone of the South Branch, but with no increase in fish kill 24 occurrence (in fact, no occurrence of fish kills). 25 26 A.2.7.3. Stressor-Response Relationship in the Field — 27 Data Used Monitored natural ammonia levels on the Rappahannock are frequently as 28 high as 0.6-0.9 mg/L. There have not been any reported kills there according to studies 29 by Steve McIninch of Virginia Commonwealth University. 30 31 Analysis None – data are actually pending.

5/2/07 79 DRAFT: DO NOT CITE OR QUOTE 1 Discussion This evidence is currently only present as a verbal communication. 2 3 Score - A large difference in response was observed relative to exposure to the 4 candidate cause, at non-spatially linked sites, but the difference is not in the expected 5 direction. The reference stream shows higher ammonia concentrations than the South 6 Branch, but no occurrence of fish kills. 7 8 A.2.7.4. Stressor-Response Relationship in the Field — 9 Data Used Mean ammonia measurements based on data from July 2004-September 10 2005, taken 5 times per month in 7 streams within West Virginia's Potomac watershed 11 and neighboring watersheds (mean = 0.089 mg N/L, max = 9.46 mg N/L)—streams that 12 did not experience kills (all WVDA sample sites) by the WV Department of Agriculture. 13 14 Analysis Mean ammonia values of 80-90 µg/L were selected from this period.

15 16 FIGURE A-24 17 A box and whisker plot of all concentrations of total ammonia nitrogen (Figures A-13 – 18 A-23). The horizontal bars are the median and 95% confidence limits. The x is the 19 mean and the box bounds the 25th-75th percentiles.

5/2/07 80 DRAFT: DO NOT CITE OR QUOTE 1 Discussion The mean level of total ammonia during the sampling year of July 2004 to 2 September 2005 was ~80-90 µg/L of ammonia. 3 The average total ammonia level during the analogous kill period from 2005 is 4 compared here to the max during the kill of 2006 under consideration. Although this is 5 an average of many sites spread throughout this part of the state, if we consider this 6 entire agricultural region as a representative reference, then we see that fish in these 7 streams are tolerating higher levels of ammonia. To account for these measurements 8 being total ammonia instead of unionized ammonia, we can estimate that under the 9 worst conditions of pHs around 9 and temperatures around 20°C, these could represent 10 unionized ammonia levels as high as 40-45 µg/L. On average though, if we look back 11 at North Fork Shenandoah data, we see that unionized ammonia levels usually 12 represent about 1/10th the amount of total ammonia levels. Using this guideline, we 13 might estimate that average unionized ammonia in all WVDA sampled streams is on the 14 order of 8-9 µg/L, which is still higher than values seen in the zone of the South Branch 15 kill during the 2006 kill. Since there was no kill during 2005, but ammonia values were 16 still comparable if not higher than in 2006, this reference does not support ammonia 17 alone being responsible for the 2006 kill. 18 19 Score - While the suitability of the entire WVDA sample set as a comparable reference 20 is not proven, fish in these waters are clearly subject to ammonia levels at least as high 21 as those seen during the 2006 kill in the South Branch with no resulting kill. 22 23 A.2.7.5. Stressor-Response Relationship from the Lab — 24 Data Used Journal articles include Milne et al. (2000) in Environmental Toxicology and 25 Chemistry, Constable et al. (2003) in Human and Ecological Risk Assessment 26 addressing Canadian fish, amphibian, and invertebrate species’ exposure to ammonia 27 and finally the U.S. EPA (1999) criteria document’s section on ammonia exposures for 28 various species. Our exposure data come from ammonia levels measured in the South 29 Branch during 2006. 30

5/2/07 81 DRAFT: DO NOT CITE OR QUOTE 1 Analysis Literature results were compared to ammonia levels reported above from the 2 South Branch. 3

4 Discussion In general, all reported NH3 thresholds for lethal or even sublethal toxic 5 effects were at least 10 times higher than what we’ve seen in the South Branch. For 6 example, Milne et al. reported needing ~400 µg/L unionized ammonia to show 7 significant lethality over a 24 hour exposure, ~750 µg/L over a 6 hour exposure, and 8 ~850 µg/L for a 1 hour exposure in rainbow trout. In longer term pulsed exposure 9 experiments of up to 6 weeks in duration, ~200 µg/L of unionized ammonia were 10 needed to cause sub-lethal effects assessed as “very sever gill damage.” Milne also 11 noted the trend that increased exposure frequency is more significant than increased 12 total concentration of ammonia in terms of effects on fish. Constable et al. reported an

13 LC50 of ~1.3 mg/L of unionized ammonia for acute exposure in fathead minnows. The

14 U.S. EPA (1999) criteria document reports EC20s of 4.79 mg/L of total ammonia at pH 8 15 for white sucker, 1.35 and 2.8 mg/L ammonia for bluegill, and a geometric mean of 4.56

16 mg/L total ammonia at pH 8 for smallmouth bass. The document also reports an LC50 17 of 50 mg/L of total ammonia during an acute exposure test in fathead minnow. The 18 highest values in the South Branch were ~2 µg/L or less of unionized ammonia during

19 the kill, reaching a maximum for the sampling period of ~170 µg/L of unionized NH3, but 20 only after the kill had stopped. It is definitely worth noting that none of these lab tests 21 exposed the fish to elevated or pulsed levels of ammonia for as long a period as our fish 22 may have been. Indeed, it would be very impractical to reproduce those conditions. 23 However, we’re also comparing the peak concentrations of ammonia observed on the 24 South Branch with reported lethal and sub-lethal levels from the literature. So even if 25 these observed levels approach those shown to cause effects in fish in the literature, we 26 should remember that ammonia levels fall far below this in between peaks and so the 27 average ammonia concentration the fish are exposed to as a sustained exposure are 28 considerably lower and definitely comparable with observed levels in reference streams 29 without kills. 30

5/2/07 82 DRAFT: DO NOT CITE OR QUOTE 1 Score If ammonia alone were responsible for the kills, we would expect the levels 2 recorded in the South Branch to be more similar to those shown to have killed fish in lab 3 tests. 4 5 A.2.8. Candidate Cause: Unspecified Toxics. 6 A.2.8.1. Spatial-Temporal Co-occurrence — 7 Data Used WVDEP Water chemistry July/August 2006 8 USGS water column parameters 9 WV Long term site (6 years) metal, nutrients, etc… 10 6 sites continuous pH/CON/DO January 2007. 11 12 Analysis NA. 13 14 Discussion Barium and arsenic were detected in the water column in 2006. No other 15 hits were observed in 2006 for toxic metals. Organics, pesticides, and pharmaceuticals 16 were not analyzed. 17 Barium in the water column but not fish tissue samples. Wirts is reviewing other 18 datasets to determine extent of the Barium problem—between 18 and 64 µg/L were 19 detected—detected at all 24/40 sites (July-August 2006 sampling index period). 20 Table A-1 presents the results of analyses of various media collected in the North 21 Fork and South Branch. The utility of the data is limited, because they are not 22 temporally and spatially relevant to the fish kill events. The table should, however, be 23 reviewed to determine similarities and differences between the two watersheds to help 24 design future studies by identifying contaminants that should be monitored. 25 26 Score No Evidence, NE. Too few toxics were analyzed in the data sets. 27 28 A.2.8.2. Evidence of Exposure or Biological Mechanisms — 29 Data Used 30 Wells fish tissue (2005) 31 Fish tissue data 32 USGS Blood plasma 33 See Table A-1. 34

5/2/07 83 DRAFT: DO NOT CITE OR QUOTE 1 Analysis NA. 2 3 Discussion Data from 2005 or earlier. West Virginia provided fish tissue data (2005) 4 verbally from Martha Wells via Pat Campbell along with fatty acid profiles (indicator of 5 how well an immune system is functioning). Sample size = 12 fish; whole fish tissue 6 analyses. Fewer lipids is taken to be an indicator of stress. The monitoring data 7 (Table 1) provide clues to be investigated with respect to future kills 8 9 Score NE, The available data do not constitute evidence that can be interpreted in 10 terms of exposure to a lethal contaminant. 11 12 A.2.8.3. Complete Causal Pathway — 13 Data Used Locations of potential sources constitute evidence of potential causal 14 pathways. 15 National Pollution Discharge Elimination System (NPDES) discharges provide evidence 16 of sources of waste water (Figure A-25). 17 Industrial chemicals used in the watershed are indicated by the Toxics Release 18 Inventory. 19 20 Discussion Sources are summarized in Table A-6. 21 National Pollution Discharge Elimination System (NPDES) discharges constitute 22 evidence of potentially toxic chemicals. Although discharge permits do not allow know 23 toxic discharges, unknown discharge constituents or unreported exceedances of 24 permitted levels may result in toxic exposures. 25 The Toxics Release Inventory data suggest other possible water pollutants. See Tables 26 A-3, A-4 and A-5 for 2005 releases. 27 The USGS document on the background of the South Branch Potomac River (USGS 28 SPMD/POCIS) (landuse, population, industry, geology, etc.) provided the information on 29 land use as an indicator of potential sources: 30 Population data for the Communities of Morefield (2500) and Petersburg (2200) 31 Romney (1940); related these sites to the fish kill locations 32 12 river miles between Petersburg and Morefield

5/2/07 84 DRAFT: DO NOT CITE OR QUOTE 1 2 FIGURE A-25 3 Locations of All Permitted Waste-water Discharges on the South Branch Potomac River 4

TABLE A-6

Potential Sources of Toxic Substances in South Branch Potomac

Source Potential Toxic stressors Poultry farms, poultry processing plants Pesticides, antibiotics, cleaners, BOD Agriculture Pesticides, metals Highways PAHs, metals POTWS Pesticides, pharmaceuticals, PAHs, metals Accidental/illegal dumping Any Golf courses Pesticides

5/2/07 85 DRAFT: DO NOT CITE OR QUOTE 1 Poultry processing plant located in the town of Morefield at mouth of South Fork 2 of South Branch; NPDES outfall within 1 mile of Morefield POTW outfall 3 Spring Run Hatchery located upstream of Petersburg on South Fork of Mill Run 4 of South Branch 5 6 Livestock numbers (cattle and chickens) have declined in the counties around the South 7 Branch PR. Chicken processing houses may ship their wastes via pretreatment plants 8 to POTW. 9 10 Score +, At least some steps of the causal pathway are present in the watershed. 11 12 A.2.8.4. Tests of Media in Laboratory 13 Data Used USEPA toxicity tests of South Branch media (Amy B.). 14 15 Analysis NA. 16 17 Discussion Toxicity tests were performed on media from the South Branch Potomac, 18 and the Cowpasture River as an undisturbed reference site. Fathead minnows were 19 tested. Results were equivocal, with some observed toxicity, but toxicity was similar in 20 the reference river as in the SB Potomac. The samples for the toxicity tests were taken 21 after the fish kills had occurred. 22 23 Score NE, No evidence, because the toxicity tests were on post-kill media. 24 25 A.2.8.5. Stressor-Response from Other Field Studies 26 Data Used NA. 27 28 Analysis NA. 29 30 Discussion Although similar fish kills have been previously observed in the other rivers 31 of the region, pathology data from those kills is not available, and no causes have been 32 identified for the other kills.

5/2/07 86 DRAFT: DO NOT CITE OR QUOTE 1 Score NE No relevant evidence. 2 3 A.2.8.6. Stressor-Response from Laboratory Studies — 4 NE, No evidence, because aqueous concentrations at the time of the kill are not 5 available for comparison to laboratory toxicity studies. 6 7 A.2.8.7. Analogous Cases — 8 Data Used None. 9 10 Analysis NA. 11 12 Discussion Prior kills in the South Branch could be analyzed as analogous cases, but 13 data for chemical exposures from those kills are lacking. 14 15 Score NE. No data from analogous kills were identified. 16 17 A.2.9. Starvation. 18 A.2.9.1. Spatial-Temporal Co-occurrence — 19 Data Used Observations of killed suckers by professional fish biologists. 20 21 Analysis NA. 22 23 Discussion Very low weight relative to length is considered the measure of starvation, 24 not just a symptom. Professional observations of dead suckers indicated that the killed 25 fish were not different from healthy fish in terms of overall size and weight. 26 27 Score - - - Convincingly weakens the case, because the candidate cause does not 28 occur. 29

5/2/07 87 DRAFT: DO NOT CITE OR QUOTE 1 A.2.9.2. Evidence of Exposure or Biological Mechanisms — 2 N E. No evidence exists on food availability or stomach contents of fish in the 2006 fish 3 kills. 4 5 A.2.9.3. Complete Causal Pathway — 6 Data Used WVDEP WAP database on benthic macroinvertebrates in South Branch. 7 8 Analysis NA. 9 10 Discussion Benthic macroinvertebrate data on the SB Potomac indicated a normal to 11 very good benthic community (WVSCI Scores 62-84) with normal abundance. 12 13 Score - Somewhat weakens, because there is at least one missing step.

5/2/07 88 DRAFT: DO NOT CITE OR QUOTE 1 APPENDIX B 2 DATA NEEDS 3 4 This appendix presents the results of discussions of data needs during the 5 workshop. 6 7 B.1. GENERAL CATEGORIES 8 B.1.1. Monitoring Data. 9 ¾ Chemical Monitoring 10 • Continuous Chemical Sampling 11 o In the Potomac, Shenandoah and reference rivers, and at 12 locations of kill events. 13 o Parameters should include pH, temp, conductivity, and TSS. 14 • Grab samples for flow events 15 o Particularly need NH3 levels in high flow, pulse events 16 • Investigate spatial variability for pH, temperature, and conductivity 17 ¾ Biological Monitoring 18 • Fish Health & Condition Data 19 o Field samples collection protocols 20 Whole fish 21 Blood 22 Select tissues 23 Field data collection recording sheet to accompany 24 o Archive protocols 25 Frozen 26 Formalin 27 Sample release protocol 28 • Only release the minimum needed to do assay 29 Data base to maintain log of samples and information about 30 the sample such as date collected, GPS coordinates, from 31 mortality event 32 o Spatial Sampling protocol 33 Including standardized reference site 34 Sample sizes for base line and kill events. 35 o Health Data 36 Behavior of organisms (location in stream) 37 • Home range (nests, etc.) 38 Gross pathology 39 Histopathology 40 • formalin-splice piece of organ 41 • healthy + sick/dead fish 42 Parasitology 43 NH3 levels in fish (blood)

5/2/07 89 DRAFT: DO NOT CITE OR QUOTE 1 • Fish population statistics 2 o General population structures 3 Periphyton 4 Snails 5 o Site selection 6 Fish kills 7 Also reference data 8 o Spatial & temporal variability 9 • Fish community assessment 10 11 B.1.2. Literature Reviews. 12 ¾ Susceptibility 13 • Mechanisms 14 o Over wintering stress 15 o Spawning stress (driven by temperature and photoperiod) 16 Smallmouth Bass 17 • (temp related, wide window, work 2 months) 18 Suckers 19 • Long spawning seasons (earlier) 20 21 B.1.3. Toxin Source Information. 22 ¾ Point and non-point sources 23 • Toxins- priority list 24 • Seasonality, pulse, etc… 25 o High volume vs. continuous 26 ¾ Agricultural practices & timing 27 ¾ TRI data 28 ¾ Water and sediment profiles 29 30 B.1.4. Laboratory Toxicity Tests (including TIE manipulations). 31 ¾ On samples collected during storm events 32 ¾ In different seasons 33 34 B.1.5. Manipulation of Exposure. 35 ¾ Field experiments are not feasible 36 37 B.1.6. Verified Predictions. 38 ¾ Periphyton: Algal changes drive changes in pH, these are light + flow limited 39 ¾ Autointoxication – Consider time lags: can be 2 weeks after change in pH for 40 auto intoxication cycle to reach build up of ammonia for to clinical effect to be 41 seen 42 43 B.1.7. Simulation Models. 44 ¾ Potentially do simulation model using Aquatox 45

5/2/07 90 DRAFT: DO NOT CITE OR QUOTE 1 B.2. CANDIDATE CAUSE-SPECIFIC DATA NEEDS 2 B.2.1. Unspecified Toxic Substances. 3 ¾ Basic work-up: flows, seasonal variations 4 ¾ Know chemical compounds list- comparisons 5 ¾ Sediment sampling 6 ¾ During spawning can sample spawning beds, include sampling finer 7 particles down in beds (habitats, nests) 8 ¾ Storm event sampling 9 • Include seasonal sampling 10 ¾ Potential sources of contaminants – “Hit List”- Upfront analysis- narrow down 11 • Water column 12 • Tissues, whole body analysis 13 o bio-accumulated chemicals 14 o non-accumulated chemicals 15 • Sediment 16 • Passive samplers 17 • Characteristic responses in fish 18 ¾ Land use – spatial relation to affected fish 19 • chemical 20 • agricultural 21 22 ¾ TRI plus MSDS- 23 • pKa’s degradation 24 • general property of chemicals 25 • total tons discharged 26 ¾ Sewage discharge (average) 27 ¾ Poultry waste amount 28 29 B.2.2. Hypoxia. 30 ¾ Blood ammonia 31 ¾ Ammonia in Water – at run off event 32 • Increases not observed in continuous monitoring. 33 • Fish can undergo the auto intoxication cycle which can delay clinical 34 effects up to 3 weeks. 35 • Criteria for reference site – temporal 36 o Fish population, geology, size, water quality, other aquatic based 37 species, invasive species 38 • pH 39 • Temperature 40 o Both air and water 41 o Cheap and continuous 42 • TSS / Turbidity Probes 43 o Conductivity

5/2/07 91 DRAFT: DO NOT CITE OR QUOTE 1 • Weather 2 o Storm Flow Events 3 • Periphyton 4 5 B.3. ROUND ROBIN-KEY PARAMETER/NEED 6 Each participant was asked to identify their highest priority data need. Data 7 needs have been roughly grouped by theme. Data needs receiving multiple votes are 8 noted in parentheses. 9 10 ¾ Total fish diagnostics (5 votes). 11 • Baseline of what’s normal, blood and tissue 12 • Samples from sites with and without kill for comparison to baseline 13 • Build archive of above 14 • Blood ammonia levels 15 • VHS diagnostic on frozen fish and new fresh samples 16 • Archive of sediment, water, fish tissue etc. samples. 17 ¾ Baseline of fish community conditions and structure (2 votes) 18 ¾ Broader geographic fish kill assessment (3 votes): 19 • When a fish kill occurs, look more broadly within the Potomac River 20 Watershed, tribs and mainstems). 21 • Spatial extent assessment of the fish kills 22 ¾ Spatial Concerns 23 • Consider combining the VA and WVA efforts—remove state 24 boundaries 25 • Centralized database of info for both WV and VA research 26 ¾ Toxins 27 • Toxic chemical hit list screen 28 • Inventory of unspecified toxics list based on TRI and historic 29 state/USGS studies plus known sources. 30 • Metal concentrations in gills 31 ¾ Temporal Concerns 32 • Continuous monitoring data: site with kills and reference sites 33 • Concurrent collection of all needed info to interpret fish kill 34 • Time windows for sampling 35 • Appropriate time data collection related to fish kills 36 • Rapid-response kits. 37 ¾ Algae 38 • Algae: Planktonic sampling for ID after storm events 39 o Community structure 40 o Spatial distribution 41 o Role in pH issue 42 o (Paul McCormick research on North Fork—periphyton research at 43 17 sites)

5/2/07 92 DRAFT: DO NOT CITE OR QUOTE 1 • Toxic algal blooms—the possibility need investigated. 2 ¾ Virus 3 • Species jump between poultry and fish—possible? 4 ¾ Other Laboratory Work 5 • Conduct laboratory exposures of pH/NH3 using fish of concern 6

5/2/07 93 DRAFT: DO NOT CITE OR QUOTE